At least four different threat actors have been identified as involved in an updated version of a massive ad fraud and residential proxy scheme called BADBOX, painting a picture of an interconnected cybercrime ecosystem.
This includes SalesTracker Group, MoYu Group, Lemon Group, and LongTV, according to new findings from the HUMAN Satori Threat Intelligence and Research team, published in

BADBOX 2.0 Botnet Infects 1 Million Android Devices for Ad Fraud and Proxy Abuse

Confidential computing strengthens application security by providing isolation, encryption and attestation so data remains protected while in use. By integrating these security features with a scalable, high-performance artificial intelligence (AI) and machine learning (ML) ecosystem, organizations can adopt a defense-in-depth approach. This is especially critical for regulated industries handling sensitive data, such as Personally Identifiable Information (PII), Protected Health Information (PHI), and financial information, enabling them to leverage AI with confidence.In this article, we expl

Confidential computing strengthens application security by providing isolation, encryption and attestation so data remains protected while in use. By integrating these security features with a scalable, high-performance artificial intelligence (AI) and machine learning (ML) ecosystem, organizations can adopt a defense-in-depth approach. This is especially critical for regulated industries handling sensitive data, such as Personally Identifiable Information (PII), Protected Health Information (PHI), and financial information, enabling them to leverage AI with confidence.

In this article, we explore how Red Hat OpenShift AI delivers a scalable, performant trusted execution environment (TEE) to deploy Kubernetes applications through confidential containers (CoCo). This is done by leveraging Azure confidential virtual machines (CVMs), powered by NVIDIA H100 GPUs with confidential computing capabilities. 

We also showcase a proof of concept (PoC) for running NVIDIA NIM™ within the CoCo environment, demonstrating an end-to-end AI inference deployment. This approach enables fine-tuned, optimized AI applications to run seamlessly on OpenShift AI, leveraging the power of NVIDIA Hopper H100 GPUs in a confidential computing framework.

This article continues previous work Red Hat and NVIDIA are collaborating on for supporting confidential GPUs as detailed in this article.   

**Red Hat OpenShift AI **

OpenShift AI is a flexible, scalable AI and ML platform that enables enterprises to create and deliver AI applications at scale across hybrid cloud environments. It’s based on the community Open Data Hub project and common open source projects such as Jupyter, Pytorch and Kubeflow.

It provides trusted, operationally consistent capabilities for teams to experiment, serve models and deliver innovative applications.

OpenShift AI also supports the creation of, training and experimentation with ML models and integrates with common AI frameworks. It includes tools for deploying AI models in production environments providing scalability, stronger security and streamlined manageability. Leveraging Red Hat OpenShift, OpenShift AI also offers infrastructure management to orchestrate containers and cloud-native infrastructure for deploying AI workloads on different footprints. It’s also an efficient way to collaborate between different company functions (data scientists, developers, IT teams) on developing, integrating and deploying models with specific business applications. 

**NVIDIA NIM **

NVIDIA NIM is a set of cloud-native microservices that simplifies and accelerates the deployment of AI models, especially for generative AI (gen AI) applications, across diverse infrastructures such as the cloud, data centres and workstations.

_NVIDIA NIM architecture_ 

NIM is an optimizing inference engine leveraging NVIDIA® TensorRT™ and NVIDIA TensorRT-LLM for low latency and high-throughput inferencing. NVIDIA NIM is built on a containerized microservice architecture including APIs, runtime layers and prebuilt model engines which are deployed on Kubernetes clusters for enhanced scalability. NVIDIA NIM supports multiple AI models such as large language models (LLMs), vision language models (VLMs), domain specific models and more. 

NVIDIA NIM can be deployed on multiple footprints including on-premises, cloud and hybrid clouds. It’s designed as an easy-to-use solution where developers can integrate AI functionality into apps while performing minimal code changes and using standard APIs. NVIDIA NIM is also a production-grade solution strengthening data security, providing observability and monitoring, autoscaling using Kubernetes and continuous security updates to production deployments. 

**Red Hat OpenShift Confidential Containers **

Red Hat OpenShift sandboxed containers, built on Kata Containers, now provide the additional capability to run **confidential containers**. Confidential containers are containers deployed within an isolated hardware enclave that help protect data and code from privileged users such as cloud or cluster administrators. The CNCF Confidential Containers project is the foundation for the OpenShift CoCo solution.

Confidential computing helps protect your data in use by leveraging dedicated hardware-based solutions. Using hardware, you can create isolated environments which are owned by you and help protect against unauthorized access or changes to your workload's data while it’s being executed (data in use).

CoCo enables cloud-native confidential computing using a number of hardware platforms and supporting technologies. CoCo aims to **standardize confidential computing at the pod level** and simplify its consumption in Kubernetes environments. By doing so, Kubernetes users can deploy CoCo workloads using their familiar workflows and tools without needing a deep understanding of the underlying confidential computing technologies.

**OpenShift AI with NVIDIA NIM and confidential containers**

Integrating NVIDIA NIM with OpenShift AI provides a unique value proposition, offering a scalable, flexible and performance-optimized platform for deploying gen AI applications. 

NVIDIA NIM's microservices architecture integrates seamlessly with OpenShift AI, creating a smooth pathway for deploying AI models within a unified workflow. This enhances consistency and simplifies management across different AI deployments. The integration also facilitates the scaling and monitoring of NVIDIA NIM deployments alongside other AI models in hybrid cloud environments. 

Incorporating CoCo into NIM and OpenShift AI improves security, provides enhanced data protection and unlocks additional avenues for AI usage in regulated industries such as financial services and healthcare.

Specifically, CoCo helps address the following Open Worldwide Application Security Project (OWASP) top 10 security issues for LLM applications, helping improve the trust and adoption of OpenShift AI with NVIDIA NIM across different industries:

*   **Prompt Injection (LLM01:2025)**: Prevents the unauthorized modification of prompts from privileged entities in the infrastructure by running inference workloads in protected and isolated environments, reducing susceptibility to manipulated inputs​.
*   **Sensitive Information Disclosure (LLM02:2025)**: Encrypting sensitive data, including cryptographic keys, AI model parameters and user data during processing reduces the risk of exposure via memory dumping attacks.
*   **Data and Model Poisoning (LLM04: 2025)**: Using a trusted execution environment minimizes data and model poisoning from compromised infrastructure or malicious privileged entities.
*   **Supply Chain Attacks on LLMs (LLM03:2025)**: Helps mitigate risks of deploying tampered or backdoored AI models through attestation and integrity checks before execution. The model can be encrypted at source and only decrypted inside the secure environment.
*   **Model Theft and Extraction (LLM03:2025)**: Protects proprietary and open source models from being extracted or replicated by making sure inferencing happens within secure and isolated environments.​
*   **Excessive Agency (LLM06:2025)**: Restricts the execution of AI agents and plugins to trusted environments so they operate within predefined policies, mitigating the impact of potential privilege escalation​ vulnerabilities.
*   **Unbounded Consumption (LLM10:2025)**: Controls execution environments to mitigate abuse of AI infrastructure, such as denial-of-service threats​ to the complete infrastructure, by restricting impact to the isolated execution environment.

**Use case: Confidential GPUs with NVIDIA NIM workloads in OpenShift AI**

Now that we have a clear understanding of what the various pieces are and what they do, let's bring them all together.

To enable a confidential GPU use case we bring together the following building blocks: 

*   **Azure public cloud**: Supporting confidential virtual machines (CVMs) used for confidential containers and providing NVIDIA H100 GPUs used for confidential GPUs
*   **OpenShift**:- The workloads' orchestration platform running on Azure
*   **OpenShift AI**: For managing the lifecycle of AI workloads (deployed via an operator)
*   **NVIDIA NIM**: For running inference using confidential GPUs (enabled via OpenShift AI)
*   **Confidential containers**: For provisioning confidential workloads connected to confidential GPUs (deployed via an operator)
*   **Attestation solution**: Based on the Trustee attestation for attesting both CPUs and GPUs in a confidential environment (deployed via an operator) 

**Architecture overview **

Let's review the solution architecture.

**TEEs, attestation and secret management**

CoCo integrates TEE infrastructure with the cloud-native world. A TEE is at the heart of a confidential computing solution, being isolated environments with enhanced security (e.g. runtime memory encryption, integrity protection) provided by confidential computing-capable hardware. The CVM that executes inside the TEE is the foundation for the OpenShift CoCo solution.

When you create a CoCo workload, a CVM is created and the workload is deployed within it. The CVM prevents anyone who isn’t the workload's rightful owner from accessing or even viewing what happens inside it.

**Attestation** is the process used to verify that a TEE, where the workload will run or where you want to send confidential information, is trusted. The combination of TEEs and attestation capability enables the CoCo solution to provide a trusted environment to run workloads and enforce the protection of code and data from unauthorized access by privileged entities.

In the CoCo solution, the Trustee project provides the capability of attestation. It’s responsible for performing the attestation operations and delivering secrets after successful attestation.

**Operators for supporting CoCo and attestation **

The Red Hat CoCo solution is based on two key operators:

*   Red Hat OpenShift confidential containers: A feature added to Red Hat OpenShift sandbox containers operator that is responsible for deploying the building blocks for connecting workloads (pods) and CVMs that run inside the TEE provided by hardware. This also enables the support for confidential GPUs.
*   Confidential compute attestation operator: Responsible for deploying and managing the Trustee service in an OpenShift cluster. Please note that the attestation operator should be deployed in a separate secure cluster the user is responsible for (more details in the following section). 

**Deploy Openshift AI and enabling NVIDIA NIM**

For our end-to-end flow to work we also need to install the following operator: 

*   Red Hat OpenShift AI operator: This deploys OpenShift AI's components
*   **NVIDIA NIM**: NVIDIA NIM is enabled from the OpenShift AI dashboard as described in the documentation. The NIM OpenShift AI application simplifies the user experience by bringing all the images available from the NVIDIA's image registry into OpenShift.

**Attesting CPU and GPU resources **

The following diagram shows how the different building blocks come together to offer confidential containers with confidential GPU capabilities: 

_Attestation for CPUs and GPUs across two OpenShift clusters (left and right)_

A key point shown in this diagram is that the confidential GPUs solution is composed of two separate OpenShift clusters: **a cluster running outside the reach of the public cloud** (where attestation is securely deployed) and **a cluster running on the public cloud** for consuming the cloud’s resources. 

**The confidential compute attestation operator** (right side), working as attester and secret resource provider, runs on the cluster outside the reach of the cloud provider. In this example, it runs on an OpenShift cluster located in a trusted environment (which the user is responsible for). 

We deploy **OpenShift sandbox containers** and **OpenShift AI** operators enabling us to create confidential containers and NVIDIA NIM workloads in the cluster running on the public cloud (left side). 

The **confidential containers** are then deployed on the OpenShift cluster (left side) with a NVIDIA NIM workload. In this PoC, we are using the NVIDIA NIM inference workload. The confidential compute attestation operator running in the trusted environment provides the required functionality to attest the CoCo environment to make sure nothing was altered by any privileged entity in the cloud environment.

For additional information on these building blocks we recommend reading our previous article: AI meets security: POC to run workloads in confidential containers using NVIDIA accelerated computing

**Workflow overview**

This illustration shows a high level workflow for deploying a model using NIM as a CoCo workload:

_High-level workflow of NIM deployed model in OpenShift AI with CoCo_

Before sending inference requests to the AI model, a user verifies the trustworthiness of the CoCo environment via remote attestation. Note that the attestation process must verify both the CPU and the GPU device.

**Demo**

**Wrap up **

With the possibility of running NVIDIA NIM microservices in a confidential TEE provided by CoCo on Openshift AI, we can accelerate the deployment of foundation models on cloud deployments and help keep confidential data secure. 

This creates opportunities for regulated industries to adopt AI inferencing use cases with a stronger security posture. Stay tuned as we work on productizing the AI application support with CoCo and follow the Red Hat confidential containers, Red Hat Openshift AI and NVIDIA NIM projects for updates.

Secure AI inferencing: POC with NVIDIA NIM on CoCo with OpenShift AI

In its latest research report, cybersecurity firm Veriti has spotted active exploitation of a vulnerability within OpenAI’s ChatGPT…

In its latest research report, cybersecurity firm Veriti has spotted active exploitation of a vulnerability within OpenAI’s ChatGPT infrastructure. Their research, shared with Hackread.com, focuses on CVE-2024-27564, a Server-Side Request Forgery (SSRF) flaw, which despite its classification as a medium-severity issue, is being actively weaponized in real-world scenarios.

Veriti’s investigation has revealed some troubling findings. One notable discovery is the very volume of attack attempts, with 10,479 attack attempts recorded in a single week originating from just one malicious IP address.

Furthermore, a significant portion of organizations, specifically 35%, are found to be inadequately protected due to misconfigurations in their security systems, including Intrusion Prevention Systems (IPS), Web Application Firewalls (WAF), and traditional firewalls.

This attack has a global impact, with the United States experiencing the highest concentration of attacks at 33%, followed by Germany and Thailand, each at 7%. Other affected regions include Indonesia, Colombia, and the United Kingdom. The following graph depicts the attack trend over time, showcasing a surge in activity during January, followed by a decrease in February and March.

(source: Veriti)

The research specifically identifies the financial sector as a primary target for these attacks. This is attributed to the heavy reliance of banks and fintech companies on AI-driven services and API integrations, which makes them susceptible to SSRF attacks that can compromise internal resources and sensitive data.

The potential consequences for financial institutions include data breaches, unauthorized transactions, regulatory penalties, and massive reputational damage. Additionally, researchers noted a critical point: “No vulnerability is too small to matter, attackers will exploit any weakness they can find.” They emphasize that overlooking medium-severity vulnerabilities can be a costly mistake, particularly for these high-value organizations.

Typically, in security practices, priority is often given to critical and high-severity vulnerabilities. However, as Veriti’s research demonstrates, attackers are opportunistic, exploiting any weakness they encounter, regardless of its severity ranking. Therefore, vulnerability prioritization should not solely rely on severity scores, as attack trends can shift rapidly, and vulnerabilities once considered insignificant can become favoured attack vectors.

Veriti has also identified a list of IP addresses actively involved in exploiting this vulnerability, providing valuable intelligence for security teams. To mitigate the risks associated with CVE-2024-27564, researchers recommend that security teams immediately review their IPS, WAF, and firewall configurations. They should also actively monitor logs for attack attempts originating from the identified malicious IP addresses and prioritize the assessment of AI-related security gaps in their risk management strategies.

Hackers Exploit ChatGPT with CVE-2024-27564, 10,000+ Attacks in a Week

Did you know that 43% of cyberattacks target small businesses, yet only 14% are prepared to defend themselves?…

Did you know that 43% of cyberattacks target small businesses, yet only 14% are prepared to defend themselves? For start-ups, the stakes are even higher. A single breach could lead to significant financial losses, tarnish your reputation and derail your growth trajectory. So, all things considered, cybersecurity is no longer a luxury or an afterthought, it is the need of the hour. 

As cybercriminals become more sophisticated, start-ups must recognise their vulnerabilities and take proactive measures to safeguard their operations. This guide explores the essential steps you need to protect your venture, from identifying risks to implementing strong cybersecurity practices.

****Why Cybersecurity Matters for Start-ups****

Start-ups often operate under the misconception that their small size makes them unlikely targets for cybercriminals. However, this couldn’t be further from the truth. Start-ups are particularly vulnerable to cyber attacks due to their limited resources, growing digital footprint, and often less-developed security infrastructure. 

Cybercriminals understand that strained resources mean that start-ups cannot invest significant time, effort or money into large-scale security measures, making them easier targets. The generally tighter-knit employee pool also means that the company is more susceptible to social engineering tactics than a large decentralised corporation. 

The consequences of a cyber attack can be devastating for a start-up. Beyond immediate financial losses, businesses face potential damage to their reputation, loss of customer trust and legal ramifications. And for many start-ups, a significant security breach could mean the difference between survival and failure.

But cybersecurity is not always about the sheer volume of resources you can throw at the problem. Understanding the nature of the attacks headed your way can significantly help you fight against them. 

****Common Cyber Risks Facing Start-ups****

_Image via Freepik_

Understanding the threats your start-up faces is the first step toward protecting against them. Here are the primary risks that start-ups need to be aware of:

****Data Breaches****

Start-ups often handle sensitive information, including customer data, intellectual property and financial records. A data breach can expose this valuable information, leading to significant financial and reputational damage. This type of attack commonly targets large-scale data storage solutions like servers or data clouds. They can be especially damaging both in terms of financial losses as well as a large hit to the company’s reputation. 

****Ransomware Attacks****

These attacks encrypt a company’s data and demand payment for its release. Start-ups are particularly vulnerable due to their often limited backup systems and immediate need for data access to maintain operations. While the MO for them vary from social engineering to physically infecting systems, it usually involves the attacker using a custom-made program to lock away access to data and demand payment with the threat of deleting or releasing said data. 

****Phishing Scams****

Sophisticated phishing attempts can trick employees into revealing sensitive information or downloading malicious software (usually by masquerading as a trusted entity). These attacks often target start-ups due to their typically less-experienced workforce. Start-ups also often consist of a small group of employees who have high-level access to sensitive information, making them more vulnerable. 

****Cloud Security Vulnerabilities****

As start-ups increasingly rely on cloud services for operations, inadequate cloud security measures can leave sensitive data exposed to unauthorised access. Solutions like large offline server farms are often not feasible from a financial standpoint for start-ups leaving them with no backup in the case of Cloud service attack. 

****Cybersecurity Best Practices for Start-ups****

Implementing strong cybersecurity measures doesn’t have to be overwhelming. Here are key practices that every start-up should adopt:

****Regular Security Assessments****

Conduct periodic security audits to identify vulnerabilities in your systems and processes. This helps you stay ahead of potential threats and address weaknesses before they can be exploited.

****Data Encryption****

Implement strong encryption protocols for all sensitive data, both in transit and at rest. This ensures that even if data is intercepted, it remains unreadable to unauthorised parties. Encryption protocols and programs can get expensive, but they are often worth the investment.

****Access Control****

Establish strict access control policies, limiting employee access to only the data and systems they need for their specific roles. This is especially a concern for start-ups which often have a very small team of employees. This hierarchical access control restricts the damage that can be inflicted, even if just one employee is compromised. 

****Backup Systems****

Maintain regular backups of all critical data and systems, storing them securely off-site or in the cloud. This provides a safety net in case of data loss or ransomware attacks.

****How to Implement Cybersecurity Measures****

Creating a strong cybersecurity foundation requires a systematic approach and ongoing commitment. Here’s how to get started:

****1\. Develop a Security Policy****

Create clear guidelines for data handling, access controls and security procedures. This policy should be documented and easily accessible to all employees.

****2\. Invest in Employee Education****

While IT professionals with qualifications like a Master of Cyber Security understand the complexities of cyber threats, most employees will need additional security training. Regular security awareness training helps employees recognise and respond to potential threats, making them your first line of defence against cyber attacks.

****3\. Implement Technical Controls****

Deploy essential security tools such as:

*   Multi-factor authentication
*   Endpoint protection solutions
*   Virtual Private Networks (VPNs)
*   Firewalls and antivirus software

****4\. Create an Incident Response Plan****

Develop and maintain a clear plan for responding to security incidents. This should include steps for containing the breach, assessing damage and notifying affected parties.

****5\. Regular Updates and Maintenance****

Keep all software and systems updated with the latest security patches. Regularly review and update security measures to address emerging threats.

****Building a Security-First Culture****

Creating a security-conscious culture is crucial for long-term success. This means making cybersecurity a priority from day one and integrating it into every aspect of your operations. Encourage open communication about security concerns and celebrate security-conscious behaviour.

****Investing in Cybersecurity****

For start-ups, cybersecurity isn’t just an IT issue – it’s a business imperative. By understanding the risks and implementing appropriate security measures, you can protect your venture’s future while building trust with customers and stakeholders. 

Remember, cybersecurity is an ongoing journey, not a destination. Start with the basics, build incrementally, and stay vigilant as your business grows.

While the initial investment in cybersecurity might seem a lot, the cost of a security breach far outweighs the resources required for prevention. By taking proactive steps to secure your start-up today, you’re investing in its long-term success and sustainability in an increasingly digital world.

Feature/Top Image via Freepik

Start-up Security 101: How to Protect Your Venture from Cybersecurity Risk

Plus: A nominee to lead CISA emerges, Elon Musk visits the NSA, a renowned crypto cracking firm’s secret (and problematic) cofounder is revealed, and more.

Knifings, firebombings, shootings, and murder-for-hire plots—all linked to a splinter group of the 764 crime network called “No Lives Matter.” According to its own manifesto, the group seeks to “purify mankind through endless attacks” and has released at least two “kill guides” tied to violent plots in the US and Europe. Intelligence documents reviewed by WIRED reveal growing concern among analysts, but experts remain unsure how to stop the group’s spread.

On Monday, X experienced intermittent outages after a botnet flooded the social network with junk traffic in an attempt to take down its system. Elon Musk stated that the distributed denial-of-service attack originated from Ukrainian IP addresses, implying that the country—already under siege by a Russian invasion and frequently mocked by the centibillionaire—may have been responsible. Security experts tell WIRED that this is not how DDoS attacks work.

Meanwhile, inside the Cybersecurity and Infrastructure Security Agency, mass layoffs are hurting US cyberdefense, weakening protections against foreign adversaries. Vital staff cuts have left employees overworked and strained international partnerships, according to interviews with staff at the agency that helps safeguard cities, businesses, and nonprofits from cyberattacks. “A lot of people are scared,” says one employee. “We’re waiting for that other shoe to drop. We don't know what's coming.” As WIRED takes you inside the agencies at the center of the uncertainty and chaos of the second Trump administration, we've updated our quick and easy guide to using Signal to help you get the most out of the messaging app’s end-to-end encryption.

That’s not all. Each week, we round up the security and privacy news we didn’t cover in depth ourselves. Click on the headlines to read the full stories. And stay safe out there.

Those “green bubbles,” the cross-platform text messages that annoy iPhone owners and keep Android users relegated to a group-chat underclass, aren’t just a cultural disconnect. They’re also a security issue: Text messages sent between Android and iOS devices—unlike blue-bubble iMessage texts or Android-to-Android messages—aren’t end-to-end encrypted, leaving them open to surveillance or interception. Now, that may finally be changing.

The GSM Association, responsible for many widely used telecom standards, this week announced that its Rich Communication Services (or RCS) protocol will now support end-to-end encryption for cross-platform texting, and Apple revealed that it will now integrate that feature of RCS into its iOS devices. Until now, Apple and Google had both supported RCS’s other features in texts sent between iOS and Android, but not end-to-end encryption, which ensures that only the devices sending and receiving messages can decrypt them and not any server or snoop that sees them in transit.

Neither Apple nor the GSMA has said exactly when the new privacy features are launching. Until then, anyone sending cross-platform messages would be wise to stick with apps like WhatsApp or Signal that have long provided end-to-end encryption—and have also helped Android and iPhone users avoid personal disputes over bubble colors.

**Sean Plankey Nominated as Head of CISA, America’s Top Cybersecurity Agency**

The White House has tapped Sean Plankey to run CISA, the agency within the Department of Homeland Security primarily responsible for American digital defense. Plankey, long considered the leading contender for the role, served in multiple cybersecurity positions in the first Trump administration and previously held senior roles in US Cyber Command. In that DOD agency focused on cyber offense, he served as a weapons and tactics branch chief and earned a Bronze Star for hacking operations in Afghanistan. CISA, like many federal agencies, has faced hundreds of personnel cuts in recent weeks, and its previous director, Chris Krebs, came under harsh criticism under the previous Trump administration for the agency’s work to counter disinformation and secure elections. Krebs was fired in a Trump tweet near the end of his term after CISA described the 2020 election, whose results Trump baselessly contested, as the “most secure in American history.”

**Elon Musk Visits the National Security Agency**

Not even the National Security Agency has been spared from Elon Musk’s scorched-earth campaign to gut the federal government. On Wednesday, The Wall Street Journal reported that Musk had visited the intelligence agency in Fort Meade, meeting with leadership to discuss staff reductions and operational changes, according to current and former US officials who spoke to the Journal.

Despite being one of the most insulated branches of US intelligence, the NSA has still found itself pulled into Musk’s orbit. The visit to Fort Meade is another sign of the sweeping nature of his influence and the extraordinary access the world’s richest man has been granted over even the most secretive federal operations.

Last month, staff at the intelligence agency received emails offering deferred resignations, signaling impending changes. Then, a week ahead of his visit, Musk publicly called for an overhaul of the intelligence and cybersecurity agency. Posting to his social media site X, Musk wrote, “The NSA needs an overhaul,” alongside an apparent agency recruitment graphic featuring a group of college-aged people of color and a list of universities where the NSA was conducting outreach—seemingly mocking the agency’s recruitment efforts.

**A Buzzy Crypto Cracking Firm Had a Hidden CoFounder Accused of Sexual Assault**

The cryptocurrency recovery firm Unciphered has made headlines with its feats of whitehat wallet cracking on behalf of customers who have lost access to their cryptocurrency fortunes. In the fall of 2023, for instance, the company cracked a model of encrypted IronKey USB drive believed to hold 7,000-plus bitcoins now worth well over half a billion dollars. Now, The Washington Post reports that one of the cofounders of that company was Morgan Marquis-Boire, a once-lauded hacker and security researcher for Google and Citizen Lab who was later accused of sexually assaulting multiple women. The reported involvement of Marquis-Boire, who has largely disappeared from the cybersecurity world after facing the sex crime accusations in 2017, was kept secret from even many of the startup’s staff, and the revelations of his role have left the company in “disarray,” according to the Post.

**A Reporter Behind an Indian Hacker-for-Hire Story Is Fighting to Regain His Citizenship**

In November of 2023, Raphael Satter was one of a team of Reuters reporters who published an in-depth feature on Appin Technology, a startup that allegedly hacked numerous celebrity and civil society targets on behalf of clients. A group with a similar name responded by suing, and obtained a court ruling that for a time successfully censored Reuters’ story—a remarkable case of an Indian judge restricting free speech internationally. Satter is now fighting a different battle following publication of that story. In late 2023, his Overseas Citizen of India (OCI) card—a kind of international citizenship extended to foreign citizens of Indian origin and those married to Indians—was revoked around the same time as the judge filed the injunction. A letter sent to Satter in December of that year accused him of “maliciously creating adverse and biased opinion against Indian institutions in the international arena.” Satter is now petitioning a Delhi court to reverse that revocation, which has prevented him from entering India to visit family there.

End-to-End Encrypted Texts Between Android and iPhone Are Coming

The UK, France, Sweden, and EU have made fresh attacks on end-to-end encryption. Some of the attacks are more “crude” than those in recent years, experts say.

Over the past decade, encrypted communication has become the norm for billions of people. Every day, Signal, iMessage, and WhatsApp keep billions of messages, photos, videos, and calls private by using end-to-end encryption by default—while Zoom, Discord, and various other services all have options to enable the protection. But despite the technology’s mainstream rise, long-standing threats to weaken encryption keep piling up.

Over the past few months, there has been a surge in government and law enforcement efforts that would effectively undermine encryption, privacy advocates and experts say, with some of the emerging threats being the most “blunt” and aggressive of those in recent memory. Officials in the UK, France, and Sweden have all made moves since the start of 2025 that could undermine or eliminate the protections of end-to-end encryption, adding to a multiyear European Union plan to scan private chats and Indian efforts that could damage encryption.

These latest assaults on encryption come as intelligence agencies and law enforcement officials in the United States have recently backtracked on years of anti-encryption attitudes and now recommend that people use encrypted communication platforms whenever they can. The drastic shift in attitude followed the China-backed Salt Typhoon hacker group’s widespread breach of major US telecoms, and it comes as the second Trump administration ramps up potential surveillance of millions of undocumented migrants living in the US. Simultaneously, the administration has been straining longtime, crucial international intelligence-sharing agreements and partnerships.

“The trend is bleak,” says Carmela Troncoso, a longtime privacy and cryptography researcher and the scientific director at the Max-Planck Institute for Security and Privacy in Germany. “We see these new policies coming up as mushrooms trying to undermine encryption.”

End-to-end encryption is designed so only the sender and receiver of messages have access to their contents—governments, tech companies, and telecom providers can’t snoop on what people are saying. Those privacy and security guarantees have made encryption a target for law enforcement and governments for decades, because officials claim that the protection makes it prohibitively difficult to investigate urgent threats such as child sexual abuse material and terrorism.

As a result, governments around the world have frequently proposed technical mechanisms to bypass encryption and allow access to messages for investigations. Cryptographers and technologists have repeatedly and definitively warned, though, that any backdoor created to access end-to-end encrypted communications could be exploited by hackers or authoritarian governments, compromising everyone’s safety. Additionally, it is likely that criminals would find ways to continue to use self-made encryption tools to conceal their messages, meaning that backdoors in mainstream products would succeed at undermining protections for the public without eliminating its use by bad actors.

Broadly, the recent threats to encryption have come in three forms, says Namrata Maheshwari, the encryption policy lead at international nonprofit Access Now. First, there are those where governments or law enforcement agencies are asking for backdoors to be built into encrypted platforms to gain “lawful access” to content. At the end of February, for example, Apple pulled its encrypted iCloud backup system, called Advanced Data Protection, from use in the UK after the country’s lawmakers reportedly hit the Cupertino company with a secret order demanding Apple provide access to encrypted files. To do so, Apple would have had to create a backdoor. The order, which has been criticized by the Trump administration, is set to be challenged in a secret court hearing on March 14.

Meanwhile, lawmakers in Sweden are also considering legislation that would require encrypted messaging companies, such as Signal and WhatsApp, to keep copies of messages that people send on their platforms so they could allow law enforcement to access suspects’ histories. Signal has said it would pull out of Sweden if the potential law goes ahead. While in France earlier this year, a proposed amendment to a drug trafficking law outlined plans to require encrypted messaging services to hand over decrypted chat messages within 72 hours of a request or face fines of up to 2 percent of annual global revenue. This week, the proposal was reportedly scrapped, while some politicians said they supported the idea.

“We’re seeing some democracies revert back to very crude approaches to circumventing encryption that we maybe thought were something of the past,” Callum Voge, the director of governmental affairs and advocacy at nonprofit Internet Society, says of efforts that could require a backdoor to be created.

In January, the head of EU law enforcement agency Europol told the Financial Times that tech companies have a “social responsibility” to provide access to encrypted messages. “Anonymity is not a fundamental right,” Catherine De Bolle told the publication. The comments expanded upon a previous statement from European police chiefs saying “we do not accept that there need be a binary choice between cybersecurity or privacy on the one hand and public safety on the other.”

The second threat, Maheshwari says, relates to an increase in proposals related to a technology known as “client-side scanning.” The process, which is sometimes called “on-device scanning,” involves scanning messages locally on a person’s device before they are encrypted, and comparing them against a database of prohibited content that is held elsewhere. Client-side scanning is an effort to contort encryption backdoors into something more palatable to privacy proponents by keeping people’s personal data on their own devices.

Ultimately, though, cryptographers and digital rights advocates have repeatedly warned that client-side scanning does not sidestep the fundamental dangers posed by creating a way for a third party to access encrypted data. The Internet Society’s Voge describes such efforts as a more “sophisticated” way that democracies have been trying to circumvent encryption in recent years.

For instance, politicians in the EU have been fiercely debating plans to scan billions of messages for potential child sexual abuse material using client-side scanning for more than three years. The unresolved debates have proved highly controversial, with multiple countries pushing to weaken encryption. “Scanning for one type of content, for instance, opens the door for bulk surveillance and could create a desire to search other encrypted messaging systems across content types,” Apple officials said in a letter first reported by WIRED in August 2023, after the company ditched its own, separate plans to introduce a form of client-side scanning on iPhones.

“It’s very divided in Europe, \[there are\] countries strongly in favor of scanning and countries strongly against it as well,” Voge says of the EU’s long-running chat monitoring plans. In May 2023, WIRED obtained leaked documents that stated many European countries’ positions on encryption. At the time, Spanish officials said they would like to prevent end-to-end encryption entirely in the EU, while many others were in favor of scanning people’s messages. Other countries, such as Germany, were against weakening encryption. Dutch political documentation says the country’s intelligence agency, AIVD, considers client-side scanning to be “too great a security risk for the digital resilience of the Netherlands.”

Finally, Maheshwari says, there is always the looming threat of potential bans or blocks for encrypted services. Toward the end of 2024, Russian officials blocked access to Signal amid the country’s ongoing full-scale war against Ukraine and widescale efforts to censor and control information environments. India has an ongoing lawsuit against WhatsApp, which could threaten its ability to operate in the country or necessitate that the platform retreat from end-to-end encryption in that market. Maheshwari also points out that while all virtual private networks do not specifically use end-to-end encryption, India has already banned multiple VPN services.

While each potential proposal that could undermine encryption is slightly different, the Internet Society’s Voge says that they’ve been met with some “stronger” pro-encryption voices coming from government or law enforcement services around the world, particularly when it comes to protecting national security.

In December, two officials from the US Cybersecurity and Infrastructure Security Agency (CISA) and the FBI, encouraged more people to use encrypted communications systems after China’s Salt Typhoon hackers gained deep access to US telecoms providers, exposing unencrypted calls and texts. “Encryption is your friend, whether it’s on text messaging or if you have the capacity to use encrypted voice communication,” one of the officials said.

Voge points out that as well as CISA and the FBI’s calls to use encrypted messaging, the Swedish armed forces has specifically cleared Signal for use with unclassified material, saying it can stop messages and calls from being intercepted by third parties.

Ahead of the UK’s March 14 legal hearing about the backdoor order reportedly made to Apple, US senators and privacy groups urged there to be more transparency about the demands and the risks to global encryption it presents. A bipartisan group of five members of Congress said the “cloak of secrecy” should be removed.

UK civil liberties groups Privacy International and Liberty also filed legal challenges over the secrecy of the proceedings. “While the UK Government seems to have come for Apple today, tomorrow it may be any other big tech companies, such as Google and Microsoft, and the day after it could be Signal, your VPN Provider, Proton and others,” Privacy International said in a statement.

Ultimately, Access Now’s Maheshwari says, efforts to defend encryption will almost certainly continue, as they have for decades, to protect people’s human rights.

“Encryption right now is exceptionally important because it's a crucial enabler of the full spectrum of human rights,” Maheshwari says. “It’s not just privacy. It is what enables you to speak freely, to exercise your freedom of expression, to organize, to assemble, to associate.”

A New Era of Attacks on Encryption Is Starting to Heat Up

February 2025 saw a record 126% surge in ransomware attacks, with Cl0p leading the charge. Hackers exploited file…

February 2025 saw a record 126% surge in ransomware attacks, with Cl0p leading the charge. Hackers exploited file transfer flaws, infostealers, and AI-driven tactics, reveals Bitdefender’s latest Threat Debrief report.

Cybersecurity just reached a new milestone; and not in a good way. According to Bitdefender’s latest Threat Debrief report, February 2025 was the worst month in history for ransomware attacks, with a 126% increase in claimed victims compared to the same period last year.

This surprising jump saw the number of victims soar from 425 in February 2024 to a staggering 962 in February 2025. The massive surge in ransomware attacks occurred despite the United States-led alliance of 40 countries, announced in November 2023, aimed at dismantling ransomware gangs and their infrastructure. The initiative focused on disrupting payments, taking down infrastructure, and enhancing intelligence sharing.

****Clop (Cl0p) Ransomware at Its Peak****

According to Bitdefender’s report shared with Hackread.com ahead of publishing on Thursday, Cl0p ransomware group Clop was responsible for more than a third of the attacks, claiming 335 victims in just one month. This makes a 300% increase from the previous month.

So, what’s behind this sudden rise in attacks? Cybersecurity experts point to a new trend that’s not so new: attackers are increasingly targeting vulnerabilities in edge network devices, such as file transfer systems and remote access tools.

Instead of focusing on specific industries, these opportunistic hackers are scanning the internet for easily exploitable flaws and launching automated attacks. For example, the Cl0p ransomware gang is notorious for exploiting vulnerabilities in MOVEit, a managed file transfer (MFT) software, with the highest frequency in 2023. The group stole so much data through MOVEit vulnerabilities that it launched a clearnet website to leak stolen information from victims worldwide.

In December 2024, Cl0p also announced exploiting security vulnerabilities in Cleo’s managed file transfer software, specifically targeting Cleo Harmony, VLTrader, and LexiCom products. Bitdefender’s Threat Debrief report also spotted Cl0p’s exploitation of Cleo vulnerabilities, especially CVE-2024-50623 and CVE-2024-55956 both rated 9.8 out of 10 in severity.

Both flaws allow attackers to execute commands remotely on compromised systems and were disclosed late last year. Despite patches being available, many organizations failed to update their systems in time, leaving them wide open to exploitation leading to the surge in victims seen in February 2025.

The illustration highlights the rapid pace at which ransomware gangs exploit vulnerabilities and shift to new targets. (Credit: Ditdefender)

****Other Notable Developments in the Ransomware World****

Beyond the record-breaking numbers, Bitdefender researchers noticed several other noteworthy trends in February 2025 including:

****FunkSec’s New Infostealer****

FunkSec, a growing ransomware group, released Wolfer, a tool designed to extract sensitive information from infected machines. It communicates with a Telegram bot to gather system details, Wi-Fi passwords, and more.

A ransomware gang using infostealers is bad news, especially as researchers recently found that cybercriminals are successfully breaching U.S. national security with infostealers as cheap as $10. Even high-security institutions like the military and the FBI have had their systems compromised, with access being sold on the dark web.

****Black Basta Gets Analyzed by AI****

On February 11, 2025, the notorious Black Basta ransomware gang had its internal chats leaked. These chats contained over 200,000 Russian-language messages. Hudson Rock’s researchers created a chatbot called BlackBastaGPT to sift through the chat logs.

Insights revealed details about their profits, use of deepfake technology, and internal conflicts. The group’s leader emphasized avoiding detection by using built-in system tools, a tactic known as “living off the land.”

****Ghost Ransomware Under Scrutiny****

A joint advisory from CISA highlighted Ghost (also known as Cring), a China-based ransomware operation exploiting older but still unpatched vulnerabilities. Recommendations include patching affected software, segmenting networks, and backing up data regularly.

****Akira’s Webcam Hack****

The Akira ransomware gang found a creative way to bypass security by hijacking a victim’s webcam. Since the device ran Linux and wasn’t monitored closely, it became the perfect launchpad for encrypting files across the network undetected.

****Top 10 Companies Most Targeted by Ransomware Gangs****

The United States, Canada, the UK, Germany, and other developed nations remain the biggest targets of ransomware groups. These countries are highly vulnerable due to their reliance on connected edge devices, cloud infrastructure, and critical business data.

In total, these are the top 10 companies most targeted by ransomware gangs:

1.  USA
2.  Canada
3.  The UK
4.  Germany
5.  France
6.  Australia
7.  Brazil
8.  Mexico
9.  Italy
10.  Sweden

For those looking to understand the full scope of modern ransomware operations and how to fight back, Bitdefender has published a comprehensive whitepaper detailing current attack methods and defence strategies. You can access it here.

Ransomware Hits Record High: 126% Surge in Attacks in February 2025

Cisco Talos’ Vulnerability Discovery & Research team recently disclosed a Miniaudio and three Adobe vulnerabilities.  
The vulnerabilities mentioned in this blog post have been patched by their respective vendors, all in adherence to Cisco’s third-party vulnerability disclosure policy.    
For Snort coverage

Thursday, March 13, 2025 14:23

Cisco Talos’ Vulnerability Discovery & Research team recently disclosed a Miniaudio and three Adobe vulnerabilities.  

The vulnerabilities mentioned in this blog post have been patched by their respective vendors, all in adherence to Cisco’s third-party vulnerability disclosure policy.    

For Snort coverage that can detect the exploitation of these vulnerabilities, download the latest rule sets from Snort.org, and our latest Vulnerability Advisories are always posted on Talos Intelligence’s website.     

**Miniaudio out-of-bounds write vulnerability **

_Discovered by Emmanuel Tacheau of Cisco Talos._   

TALOS-2024-2063 (CVE-2024-41147) is an out-of-bounds write vulnerability in Miniaudio, a lightweight, single-file audio playback and capture library written in C. A missing allocation size check can cause a buffer overflow, leading to this out-of-bounds write. This vulnerability can be triggered by a specially crafted FLAC file, resulting in a memory corruption when in playback mode. The application sends raw audio data to Miniaudio, which is then played back through the default playback device as defined by the operating system. 

**Adobe Acrobat out-of-bounds write vulnerability **

_Discovered by KPC of Cisco Talos._   

TALOS-2025-2134 (CVE-2025-27163) and TALOS-2025-2136 (CVE-2025-27164) are out-of-bounds read vulnerabilities in the font functionality, which can lead to disclosure of sensitive information. TALOS-2025-2135 (CVE-2025-27158) is a memory corruption vulnerability, stemming from an uninitialized pointer in the font functionality of Adobe Acrobat, which can potentially lead to arbitrary code execution. A specially crafted font file embedded into a PDF can trigger these vulnerabilities. An attacker needs to trick the user into opening a malicious file.

Miniaudio and Adobe Acrobat Reader vulnerabilities

Cary, North Carolina, 13th March 2025, CyberNewsWire

**Cary, North Carolina, March 13th, 2025, CyberNewsWire**

As Artificial Intelligence (AI)-powered cyber threats surge, INE Security, a global leader in cybersecurity training and certification, is launching a new initiative to help organizations rethink cybersecurity training and workforce development. The company warns that AI is reshaping both the threat landscape and the skills required for cybersecurity professionals. While AI offers significant advantages in cyber defense, organizations must ensure their teams are properly trained to leverage it effectively without becoming overly reliant on automation.

> “The rise of AI in cybersecurity isn’t just a challenge—it’s an opportunity,” said Dara Warn, CEO of INE Security. “By training cybersecurity professionals properly, AI can be leveraged to filter noise, reduce burnout, and increase efficiency. However, if we don’t train people to understand the ‘why’ behind AI-driven decisions, we risk a future where cybersecurity professionals are blindly following AI without the expertise to think critically beyond it.”

**AI as a Force Multiplier: Improving SOC Efficiency and Threat Detection**

AI-driven security tools are improving the signal-to-noise ratio, making Security Operations Centers (SOCs) more efficient by reducing false positive alerts—an area cybersecurity tools have been refining for over a decade. AI can prioritize critical threats, allowing analysts to focus on real dangers rather than wasting time investigating false alarms.

> “AI is making threat detection smarter, but it’s not foolproof,” said Tracy Wallace, Director of Content at INE Security. “Security professionals need to be trained to work alongside AI, not just follow its outputs. AI is great at reducing alert fatigue, but analysts still need the expertise to investigate, interpret, and respond to threats accurately.”

**Generative AI: A Double-Edged Sword for Cybersecurity Talent**

One of the most promising yet complex aspects of AI’s rise is its impact on the cybersecurity workforce. On one hand, generative AI will lower the barrier to entry, allowing more professionals to enter the cybersecurity field and reducing the global labor shortage. 

> However, this shift also presents risks. “The concern isn’t that AI is making cybersecurity easier,” said Wallace. “The concern is that if professionals become too dependent on AI outputs, they won’t develop the critical-thinking skills necessary to work beyond what the AI gives them. Organizations must ensure that cybersecurity training teaches professionals not just how to use AI but how to work independently of it when needed.”

**The Data Privacy Dilemma: AI and LLM Security Risks**

Another concern in AI-driven cybersecurity is data privacy and security risks with large language models (LLMs). While concerns over data leakage with cloud-based AI models are growing, this isn’t a new challenge—it’s an evolution of longstanding security principles. Organizations must ensure AI-powered security solutions do not require external data sharing.

> “As AI becomes more deeply integrated into cybersecurity operations, privacy-first security architectures are crucial,” said Wallace. “Organizations need AI models that can operate securely without exposing sensitive data to external systems.”

**The Future of AI Security Training: Agentic Architectures and AI-Driven Automation**

Looking ahead, Agentic AI architectures are becoming a hot topic in cybersecurity. While some view it as buzzword hype, there is real potential for AI-driven security agents that autonomously investigate threats, adjust defenses in real-time, and improve security workflows with minimal human intervention.

> However, automation must be carefully balanced. “Agentic AI might be the future, but we can’t let it replace hands-on expertise and human decision-making,” said Warn. “Security professionals must be trained to interpret AI-driven insights, make judgment calls, and recognize when AI is wrong.”

**Training as the Solution: INE Security’s AI-Powered Cybersecurity Curriculum**

To close the cybersecurity skills gap and help professionals work effectively with AI, INE Security is working to expand its AI-driven training programs. These programs will focus on:

*   **AI-Driven Threat Analysis** – Training security teams to interpret AI-generated threat intelligence and reduce false positives.
*   **Machine Learning for Cyber Defense** – Teaching professionals how AI-powered security models work and how attackers exploit AI vulnerabilities.
*   **Generative AI in Cybersecurity** – Helping cybersecurity teams understand the risks and benefits of AI-generated attacks and defenses.
*   **Hands-On AI Security Labs** – Simulating real-world AI-powered attacks and training professionals on how to counter them manually and with AI assistance.

> “Our end goal is not just to train security professionals how to use AI but to train them how to think critically in an AI-driven world,” said Wallace. 

**The Call to Action: Prepare for AI-Driven Threats Now**

With AI transforming cybersecurity threats at an unprecedented pace, INE Security urges companies to:

*   Train their cybersecurity teams on AI-driven tools, while ensuring they develop critical problem-solving skills.
*   Prioritize AI-powered security solutions that enhance, not replace, human expertise.
*   Implement privacy-first AI models that reduce data exposure risks**.**

> “The AI revolution in cybersecurity is here,” concluded Warn. “Organizations that act now—by investing in security training, developing cybersecurity talent, and understanding how AI truly impacts the field—will be the ones leading the industry forward. The future of cybersecurity belongs to those who train for it.”

**About INE Security**

INE Security is the premier provider of online networking and cybersecurity training and certification. Harnessing a powerful hands-on lab platform, cutting-edge technology, a global video distribution network, and world-class instructors, INE Security is the top training choice for Fortune 500 companies worldwide for cybersecurity training in business and for IT professionals looking to advance their careers, offering both Red Team training and Blue Team training. INE Security’s suite of learning paths offers an incomparable depth of expertise across cybersecurity and is committed to delivering advanced technical training while also lowering the barriers worldwide for those looking to enter and excel in an IT career.

**Contact**

**Kathryn Brown**  
**INE Security**  
**\[email protected\]**

INE Security Alert: Using AI-Driven Cybersecurity Training to Counter Emerging Threats

Employees at the Cybersecurity and Infrastructure Security Agency tell WIRED they’re struggling to protect the US while the administration dismisses their colleagues and poisons their partnerships.

Mass layoffs and weak leadership are taking a severe toll on the US government’s cyber defense agency, undermining its ability to protect America from foreign adversaries bent on crippling infrastructure and ransomware gangs that are bleeding small businesses dry.

Inside the Cybersecurity and Infrastructure Security Agency, vital support staff are gone, international partnerships have been strained, and workers are afraid to discuss threats to democracy that they’re now prohibited from countering. Employees are even more overworked than usual, and new assignments from the administration are interfering with important tasks. Meanwhile, CISA’s temporary leader is doing everything she can to appease President Donald Trump, infuriating employees who say she’s out of touch and refusing to protect them.

“You've got a lot of people who … are looking over their shoulder as opposed to looking at the enemy right now,” says one CISA employee.

As the Trump administration’s war on the federal bureaucracy throws key agencies into chaos, CISA’s turmoil could have underappreciated consequences for national security and economic prospects. The agency, part of the Department of Homeland Security, has steadily built a reputation as a nonpartisan source of funding, guidance, and even direct defensive support for cities, businesses, and nonprofits reeling from cyberattacks. That mission is now under threat, according to interviews with seven CISA employees and another person familiar with the matter, all of whom requested anonymity to avoid reprisals.

“Our enemies are not slowing their continuous assaults on our systems,” says Suzanne Spaulding, who led CISA’s predecessor during the Obama administration. “We need all hands on deck and focused, not traumatized and distracted.”

**Talent Exodus**

CISA’s mission has grown significantly since its creation in 2018. Established mainly to defend government networks, the agency increasingly embraced new roles supporting private companies and state governments, advocating for secure software, and cooperating with foreign partners. This helped CISA raise its profile and gain credibility. But now, following several rounds of layoffs and new restrictions from the Trump administration, the agency is struggling to sustain its momentum.

The extent of the cuts at CISA is still unclear—employees are only learning about the loss of colleagues through word of mouth—but multiple employees estimate that, between the layoffs and the Office of Personnel Management’s deferred-resignation program, CISA has lost between 300 and 400 staffers—roughly 10 percent of its 3,200-person workforce. Many of those people were hired through DHS’s Cybersecurity Talent Management System (CTMS), a program designed to recruit experts by competing with private-sector salaries. As a result, they were classified as probationary employees for three years, making them vulnerable to layoffs. These layoffs at CISA also hit longtime government workers who had become probationary by transferring into CTMS roles.

Key employees who have left include Kelly Shaw, who oversaw one of CISA’s marquee programs, a voluntary threat-detection service for critical infrastructure operators; David Carroll, who led the Mission Engineering Division, the agency’s technological backbone; and Carroll’s technical director, Duncan McCaskill. “We've had a very large brain drain,” an employee says.

The departures have strained a workforce that was already stretched thin. “We were running into \[a\] critical skills shortage previously,” says a second employee. “Most people are and have been doing the work of two or more full-time \[staffers\].”

The CISA team that helps critical infrastructure operators respond to hacks has been understaffed for years. The agency added support positions for that team after a Government Accountability Office audit, but “most of those people got terminated,” a third employee says.

CISA’s flagship programs have been mostly unscathed so far. That includes the threat-hunting branch, which analyzes threats, searches government networks for intruders, and responds to breaches. But some of the laid-off staffers provided crucial “backend” support for threat hunters and other analysts. “There's enhancements that could be made to the tools that they're using,” the first employee says. But with fewer people developing those improvements, “we're going to start having antiquated systems.”

In a statement, DHS spokesperson Tricia McLaughlin says CISA remains “committed to the safety and security of the nation’s critical infrastructure” and touted “the critical skills that CISA experts bring to the fight every day.”

National Security Council spokesperson James Hewitt says the reporting in this story is “nonsense,” adding that “there have been no widespread layoffs at CISA and its mission remains fully intact.”

“We continue to strengthen cybersecurity partnerships, advance AI and open-source security, and protect election integrity,” Hewitt says. “Under President Trump’s leadership, our administration will make significant strides in enhancing national cybersecurity.”

**Partnership Problems**

CISA’s external partnerships—the cornerstone of its effort to understand and counter evolving threats—have been especially hard-hit.

International travel has been frozen, two employees say, with trips—and even online communications with foreign partners—requiring high-level approvals. That has hampered CISA’s collaboration with other cyber agencies, including those of “Five Eyes” allies Canada, Australia, New Zealand, and the UK, staffers say.

CISA employees can’t even communicate with people at other federal agencies the way they used to. Previously routine conversations between CISA staffers and high-level officials elsewhere now need special permissions, slowing down important work. “I can’t reach out to a CISO about an emergency situation without approval,” a fourth employee says.

Meanwhile, companies have expressed fears about sharing information with CISA and even using the agency’s free attack-monitoring services due to DOGE’s ransacking of agency computers, according to two employees. “There is advanced concern about all of our services that collect sensitive data,” the third employee says. “Partners \[are\] asking questions about what DOGE can get access to and expressing concern that their sensitive information is in their hands.”

“The wrecking of preestablished relationships will be something that will have long-lasting effects,” the fourth employee says.

CISA’s Joint Cyber Defense Collaborative, a high-profile hub of government-industry cooperation, is also struggling. The JCDC currently works with more than 300 private companies to exchange threat information, draft defensive playbooks, discuss geopolitical challenges, and publish advisories. The unit wants to add hundreds more partners, but it has “had difficulty scaling this,” the first employee says, and recent layoffs have only made things worse. Contractors might be able to help, but the JCDC’s “vendor support contracts run out in less than a year,” the employee says, and as processes across the government have been frozen or paused in recent weeks, CISA doesn’t know if it can pursue new agreements. The JCDC doesn't have enough federal workers to pick up the slack, the fourth CISA employee says.

With fewer staffers to manage its relationships, the JCDC confronts a perilous question: How should it focus its resources without jeopardizing important visibility into the threat landscape? Emphasizing ties with major companies might be more economical, but that would risk overlooking mid-sized firms whose technology is quietly essential to vital US industries.

“CISA continuously evaluates how it works with partners,” McLaughlin says, “and has taken decisive action to maximize impact while being good stewards of taxpayer dollars and aligning with Administration priorities and our authorities.”

**Gutting Security Advocacy**

Other parts of CISA’s mission have also begun to atrophy.

During the Biden administration, CISA vowed to help the tech industry understand and mitigate the risks of open-source software, which is often poorly maintained and has repeatedly been exploited by hackers. But since Trump took office, CISA has lost the three technical luminaries who oversaw that work: Jack Cable, Aeva Black, and Tim Pepper. Open-source security remains a major challenge, but CISA’s efforts to address that challenge are now rudderless.

The new administration has also frozen CISA’s work on artificial intelligence. The agency had been researching ways to use AI for vulnerability detection and networking monitoring, as well as partnering with the private sector to study AI risks. “About 50 percent of \[CISA’s\] AI expert headcount has been let go,” says a person familiar with the matter, which is “severely limiting” CISA’s ability to help the US Artificial Intelligence Safety Institute test AI models before deployment.

The administration also pushed out CISA’s chief AI officer, Lisa Einstein, and closed down her office, the person familiar with the matter says. Einstein’s team oversaw CISA’s use of AI and worked with private companies and foreign governments on AI security.

A large team of DHS and CISA AI staffers was set to accompany Vice President JD Vance to Paris in February for an AI summit, but those experts “were all pulled back” from attending, according to a person familiar with the matter.

**‘Nefarious’ Retribution**

CISA staffers are still reeling from the agency’s suspension of its election security program and the layoffs of most people who worked on that mission. The election security initiative, through which CISA provided free services and guidance to state and local officials and worked with tech companies to track online misinformation, became a target of right-wing conspiracy theories in 2020, which marked it for death after Trump’s return to the White House.

The program—on hold pending CISA’s review of a recently completed internal assessment—was a tiny part of CISA’s budget and operations, but the campaign against it has alarmed agency employees. “This is definitely in the freak-out zone,” says the first employee, who adds that CISA staffers across the political spectrum support the agency’s efforts to track online misinformation campaigns. “All of us recognize that this is a common deception tactic of the enemy.”

The election security purge rippled across the agency, because some of the laid-off staffers had moved from elections to other assignments or were simultaneously working on both missions. Geoff Hale, who led the elections team between 2018 and 2024, was serving as the chief of partnerships at the JCDC when he was placed on administrative leave, setting off a scramble to replace him.

The removal of Hale and his colleagues “was the start to a decline in morale” at CISA, according to the second employee. Now, staffers are afraid to discuss certain topics in public forums: “No one's going to talk about election security right now,” the first employee says.

“The fact that there’s retribution from the president … is kind of frightening,” this employee adds. “A very nefarious place to be.”

**Abandoned and Demoralized**

The layoffs, operational changes, and other disruptions at CISA have severely depleted morale and undermined the agency’s effectiveness. “Even simple tasks feel hard to accomplish because you don't know if your teammates won't be here tomorrow,” says the fourth employee.

The biggest source of stress and frustration is acting CISA director Bridget Bean, a former Trump appointee who, employees say, appears eager to please the president even if it means not defending her agency. Bean “just takes whatever comes down and implements \[it\] without thought of how it will affect \[CISA’s\] mission,” the fifth employee says. Employees describe her as a poor leader and ineffective communicator who has zealously enacted Trump’s agenda. In town-hall meetings with employees, Bean has said CISA must carefully review its its authorities and urged staffers to “assume noble intent” when dealing with Trump officials. While discussing Elon Musk’s mass-buyout program, she allegedly said, “I like to say ‘Fork in the Road’ because it's kind of fun,.” according to the fourth employee. She was so eager to comply with Musk’s “What did you do last week?” email that she instructed staffers to respond to it before DHS had finalized its department-wide approach. DHS later told staff not to respond, and Bean had to walk back her directive.

“Bean feels like she's against the workforce just to please the current administration,” the second employee says. The fourth employee describes her as “not authentic, tone-deaf, spineless, \[and\] devoid of leadership.”

McLaughlin, the DHS spokesperson, says CISA “is not interested in ad hominem attacks against its leadership,” which she says “has doubled down on openness and transparency with the workforce.”

The return-to-office mandate has also caused problems. With all employees on-site, there isn’t enough room in CISA’s offices for the contractors who support the agency’s staff. That has made it “very difficult” to collaborate on projects and hold technical discussions, according to the first employee. “There wasn’t much thought about \[RTO’s\] impact to operations,” says the fourth employee. According to a fifth employee, “executing some of our sensitive operations is now harder.” (“CISA has worked tirelessly to make the return to office as smooth as possible from space to technology,” McLaughlin says.)

Employees are dealing with other stressors, too. They have no idea who’s reading their Musk-mandated performance reports, how they’re being evaluated, or whether AI is analyzing them for future layoffs. And there’s a lot of new paperwork. “The amount of extra shit we have to do to comply with the ‘efficiency measures’ … \[takes\] a lot of time away from doing our job,” says the fifth employee.

**Bracing for More**

When Trump signed the bill creating CISA in November 2018, he said the agency’s workforce would be “on the front lines of our cyber defense” and “make us, I think, much more effective.” Six and a half years later, many CISA employees see Trump as the biggest thing holding them back.

“This administration has declared psychological warfare on this workforce,” the fourth employee says.

With CISA drawing up plans for even larger cuts, staffers know the chaos is far from over.

“A lot of people are scared,” says the first employee. “We’re waiting for that other shoe to drop. We don't know what's coming.”

Entire wings of CISA—like National Risk Management Center and the Stakeholder Engagement Division—could be on the chopping block. Even in offices that survive, some of the government’s most talented cyber experts—people who chose public service over huge sums of money and craved the lifestyle of CISA’s now-eliminated remote-work environment—are starting to see their employment calculus differently. Some of them will likely leave for stabler jobs, further jeopardizing CISA’s mission. “What is the organization going to be capable of doing in the future?” the first employee asks.

If Trump’s confrontational foreign-policy strategy escalates tensions with Russia, China, Iran, or North Korea, it’s likely those nations could step up their use of cyberattacks to exact revenge. In that environment, warns Nitin Natarajan, CISA’s deputy director during the Biden administration, weakening the agency could prove very dangerous.

“Cuts to CISA’s cyber mission,” Natarajan says, “will only negatively impact our ability to not only protect federal government networks, but those around the nation that Americans depend on every day.”

Tag

#intel