Google AI Uncovers 13-Year-Old Chrome Flaw Amid Record Patching Pace

Google this week confirmed that this year’s surge in Chrome vulnerabilities has been driven by the use of AI.

The increase in Chrome bugs started in April and has continued throughout July, with the latest browser release containing 370 security fixes and bringing the total number of bugs patched this year to over 1,800.

Google patched 1,072 security defects with the Chrome 149 and 150 releases, “surpassing the total number of security bugs fixed across the prior 23 milestones combined.”

This was the result of using an agent harness that leverages Gemini to identify security flaws across Chrome’s codebase with increased efficiency, the internet giant says.

What validated the AI-powered vulnerability detection, Google says, was the discovery of a sandbox escape that lurked in the browser for 13 years.

Tracked as CVE-2026-3545 (CVSS score of 9.8) and patched in early May in Chrome 145, the issue could have allowed “a compromised renderer to trick the browser into reading local files,” it says.

Advertisement. Scroll to continue reading.

Described as an insufficient data validation in Navigation, the vulnerability could have been exploited via crafted HTML pages to perform a sandbox escape.

The Chrome Security team started using LLMs in 2023, but it was the agent harness built in early 2026 that has been essential to finding and patching more flaws in the browser.

It has support for model interoperability, was trained on a knowledge base of previously identified CVEs and of Chrome’s entire Git history, leverages developer-supplied SECURITY.md files that are consumed by a “critic” agent, and can run vulnerability-finding models over the codebase multiple times.

“We’ve built all of this with safety in mind, and have put in place guardrails to mitigate the risk of AI behaving unexpectedly. Our AI analyzes source code strictly at rest, operating on locked-down machines that lack general internet access,” Google explains.

While also relying on AI for bug validation, triaging, and patch generation, the internet giant continues to use other security testing infrastructure for bug detection, and welcomes researchers submitting vulnerability reports through its VRP.

“At this point, we have LLMs generating candidate fixes for most vulnerabilities, dramatically increasing the rate of security fixes in recent Chrome releases,” Google says.

Additionally, the company is relying on AI to prevent new security bugs by identifying and eliminating them “as close to code submit-time as possible” and by neutralizing issues in the tree that could impact code deemed safe in isolation.

Reducing the patch gap and eliminating memory safety bugs

To reduce the patch gap, Google is also piloting a twice-a-week Chrome security release cadence, which will complement the two-week schedule for major Chrome releases that was announced in March.

“We are working on automating the generation of release notes and CVE descriptions from security bug fixes to eliminate manual bottlenecks and shorten the window between vulnerability discovery and public disclosure,” the company says.

Google is also working on ensuring that Chrome is always up to date on users’ machines, through dynamic patching, which eliminates the need to restart the browser when applying updates, seamless session restoration, and restarts while in a windowless state (the application runs in the background while all windows are closed).

Additionally, Google is working on eliminating entire classes of security issues from Chrome, such as memory safety flaws. For that, it is hardening the runtime environment to squash C++ bugs, and is transitioning to memory-safe languages.

In this regard, it is expanding MiraclePtr to more libraries and deploying MiracleObject to the GPU main thread to neutralize use-after-free defects. In addition, it’s taking a ‘spanification’ effort to eliminate out-of-bounds weaknesses, and is integrating memory allocation protections to block integer overflows.

In the long term, Google is transitioning Chrome’s codebase to memory-safe languages like Rust, through building a centralized Rust SDK, deploying Rust to replace most flawed code segments, and writing new modular components in Rust.

“In addition to Rust, we are also exploring options like implementing the browser’s top-level user interface using HTML, CSS, and TypeScript to further reduce dependencies on traditional C++ frameworks,” Google notes.

According to the company, keeping the open source software ecosystem secure is another critical aspect of improving Chrome users’ protections. It has been using automated vulnerability scanning pipelines to keep dependencies patched and will move all Chrome third-party dependencies onto automated update pipelines.

“Every bug found and fixed is one less foothold for an attacker. But discovering and fixing a bug is only half the battle — we must also ship the fix and apply the update for users faster than adversaries can exploit the bug, and invest in projects that mitigate or eliminate classes of bugs through accelerated release cadences, dynamic patching, and opportune restarts, we are driving toward a browser that is continuously protected without disrupting the user,” Google notes.

https://www.securityweek.com/googles-ai-agent-uncovers-13-year-old-chrome-flaw-amid-record-patching-pace/

EU to Crack Down on AI Deepfakes, Illicit Imagery and Hacking With New Team in Brussels

The European Union rolled out a new team on Friday to rein in AI companies across the world, in one of the most aggressive regulations the high-tech sector has so far faced as fears rise over the risks the rapidly advancing technology poses to people, politics and prosperity.

Brussels aims to track the use of AI models for violations of its new regulations, like the publishing of sexually explicit material, fake photos and videos, and cyber threats to public infrastructure. When the bloc’s AI Act comes into force on Sunday, AI companies will be required to make clear to consumers with labels or digital watermarks that chatbots or imagery are generated with AI.

“As enforcement begins, we are taking an important step towards AI that people and businesses can understand and trust, and whose benefits are shared widely across our society,” said Henna Virkkunen, the EU chief for tech sovereignty, on Friday.

The European Commission said in a statement that new regulations also include “systemic risks” posed by AI like “chemical, biological, radiological and nuclear incidents, loss of control, cyber offense, harmful manipulation and threats to fundamental rights.”

The team is the latest move in the 27-nation EU’s “tech sovereignty” strategy that welds landmark digital regulations with economic ambition that has seen over the past week billions of euros in fines on Big Tech companies as well as record investment in AI infrastructure inside the bloc.

The rollout comes on the heels of shocking AI safety failures that rattled the nascent industry. Political leaders worldwide are now weighing control of the technology with market leadership.

Advertisement. Scroll to continue reading.

Anthropic said its artificial intelligence models hacked into three other organizations during testing, just days after ChatGPT maker OpenAI raised concerns over AI controls after it disclosed its rogue models hacked another company.

The EU is now expanding its AI Office in Brussels with an additional 38 people who will begin monitoring AI companies, from the new firms to the American and Chinese tech titans like OpenAI and DeepSeek.

Those companies must “document certain information,” said the European Commission, the bloc’s tech enforcer, which reserves the right to interview AI company staff during investigations. It has also launched a Whistleblower Tool for tech workers and a Compliance Tool for tech users to confidentially alert authorities to illegal conduct.

If models or other products break the EU’s sector regulations, called the AI Act, Brussels can fine the firms or cut off their access to the EU market. Recent gigantic antitrust fines on U.S. tech companies have irked U.S. President Donald Trump.

The EU now clearly sees systemic vulnerability in its deep reliance on American software companies like Amazon, Google and Microsoft as well as imports of Chinese industrial goods and critical minerals. While it is seeking protections from AI, it is also keen to catch up in the AI arms race, where it is a distant third behind the U.S. and China.

More broadly, the EU is seeking more independence from both Washington and Beijing by reinvigorating specific domestic industries like manufacturing and defense and forging new trade deals from Brazil to Australia to meet the global rise in economic nationalism spearheaded by Trump.

https://www.securityweek.com/eu-to-crack-down-on-ai-deepfakes-illicit-imagery-and-hacking-with-new-team-in-brussels/

Prompted by OpenAI Disclosure, Anthropic Finds Its Own Models Hacked 3 Organizations

Anthropic revealed on Thursday that some of its Claude models escaped test environments and hacked into the systems of three organizations while trying to solve a challenge.

The admission comes after OpenAI revealed that its models broke out of what was believed to be an isolated environment and breached Hugging Face and other organizations.

Anthropic decided to conduct its own investigation after the OpenAI incident came to light, reviewing 141,000 evaluation runs where Claude could have gained internet access.

The analysis revealed three instances where a model reached the public web either from within or while interacting with an environment set up by Irregular, an Israeli AI security startup that serves as one of Anthropic’s third-party evaluation partners.

The models that broke out from the testing environment then breached the production systems of three unnamed organizations. The earliest intrusion dates to April, but none of the targeted organizations detected the attacks on their own.

According to the AI giant, Claude had been tasked with conducting a capture-the-flag challenge to assess its cyber capabilities.

Advertisement. Scroll to continue reading.

The breakout from the sandbox occurred due to a misunderstanding between Anthropic and Irregular—Anthropic told Claude that it was conducting a simulation in an environment without internet, but a connection to the internet was in fact available and the models treated it as part of the exercise. The AI then also falsely believed that the organizations it targeted were part of the exercise.

Anthropic claims this is the only reason why its models conducted real-world attacks, and it was never their intention to escape the test environment.

Unlike in the OpenAI incident, which involved the exploitation of a JFrog product zero-day, the hacking conducted by Claude models leveraged weak credentials, unauthenticated endpoints, and other basic attack techniques.

The real-world attacks

Anthropic said the incidents involved Mythos, Opus, and an internal research model, all running without the safeguards deployed when they are made available to customers. The internal model, which the company described as its “latest”, stopped the exercise when it realized that it had breached a real company.

In the first incident, Claude Opus 4.7 continued to attack the system even after learning that it was operating in a real environment rather than a test environment, believing that the hacked company was part of the evaluation. The model targeted the company because its website domain name resembled the name of the fictional target company set up for the test.

In the second incident, Mythos 5 gained access to the systems of a cybersecurity company after deploying a malicious Python package to PyPI. The security firm regularly scans Python packages and it installed the malicious package, which enabled the AI to exfiltrate credentials and access the company’s infrastructure.

This incident demonstrates the complexity of the actions AI models can carry out. As described by Anthropic:

[In] order to create a PyPI account, Claude needed an email address. And in order to create an email address, it needed a phone number. To get a phone number, after failing to find a free phone number service, it tried—and failed—to obtain funds to pay for a phone number through several different means. It finally backtracked, found a free, non-blocked email provider, used this to register a PyPI account, and then used this account to upload malware to PyPI.

The third intrusion was conducted by the internal model, which stopped operating when it realized that the systems it was accessing were no longer part of the capture-the-flag challenge, but not before using exposed credentials and SQL injection flaws to compromise a company’s internet-facing app.

Anthropic concluded this was primarily a harness and operational failure rather than a case of models pursuing their own goals or deliberately deceiving evaluators.

The company said the incident underscores the need for stricter internet-isolation verification and containment controls in third-party testing environments, and it’s encouraging other AI labs to conduct similar reviews of their own cybersecurity evaluations.

https://www.securityweek.com/after-openai-disclosure-anthropic-finds-its-own-models-hacked-3-organizations/

Critical Flaw Allowed to Azure Cosmos DB Pwnage

A critical vulnerability in the Azure Cosmos DB database service could have allowed attackers to compromise all databases on the service, cybersecurity outfit Wiz reports.

Referred to as CosmosEscape, the security defect could have allowed an attacker to obtain a platform-wide key and retrieve the primary key of any Cosmos DB account, gaining full read and write access across the service.

Armed with the key, an attacker could have listed all databases on the service, filtering them by specific organization identifiers, such as subscription and tenant IDs.

“Chained together, these capabilities could have enabled precision targeting at platform scale: from identifying a specific organization’s databases to compromising them, all from publicly accessible endpoints,” Wiz says.

According to the cybersecurity firm, because Microsoft uses Cosmos DB to store data across Entra ID, Teams, and Copilot, the vulnerability potentially exposed the tech giant’s databases to unauthorized access.

The issue could have been exploited through the Gremlin API, a popular graph query language that relies on a custom Gremlin engine, compiling queries into .NET code that was executed in a sandbox.

Advertisement. Scroll to continue reading.

Pwning the sandbox

While the engine enforced restrictions to prevent access beyond Gremlin operations, it did not account for .NET reflection, which Wiz used to build arbitrary code execution primitives.

“By bypassing the Gremlin sandbox, we’ve gained code execution on the DB Gateway, a service that executes customer queries on their behalf, running on multi-tenant Service Fabric clusters,” Wiz explains.

The cybersecurity firm then discovered that the gateway was using a signing key to retrieve customer accounts’ primary keys and access their databases, and that the signing key worked across tenants, regions, and even APIs.

“It was a platform-wide key that could retrieve the primary key for any Cosmos DB account on the service, all through publicly accessible endpoints. We dubbed it the Cosmos Master Key,” Wiz notes.

Through the master key, the cybersecurity firm’s researchers could access a configuration store containing details on every Cosmos DB account, including names, subscription IDs, tenant IDs, and various other configuration data.

Pwning the databases

“The Config Store was itself a Cosmos DB database, meaning it could be queried with the full flexibility of CosmosDB’s SQL engine. It also meant that the Cosmos Master Key could retrieve its primary key – enabling attackers that exploited CosmosEscape to list all accounts in a region, or query by specific tenant ID to identify a specific organization’s databases,” the company explains.

An attacker with the master key and access to the configuration store could enumerate all Cosmos DB accounts, filter them, and retrieve a target’s primary key to gain full read and write access to all their databases.

According to Wiz, the attack could also be mounted against private and network-isolated Cosmos DB accounts, as well as Microsoft’s own databases.

Wiz reported the vulnerability to Microsoft in November 2025. Within two days, Microsoft deployed a hotfix to block the attack vector. In July, the tech giant completed the rollout of a long-term architectural fix across all regions.

“Microsoft conducted extensive reviews of access logs and found no evidence of unauthorized activity outside of the researcher’s testing activity, and no customer data was accessed. There is no customer action required,” the company said.

https://www.securityweek.com/critical-flaw-led-to-azure-cosmos-db-pwnage/

Product Spotlight on Mobile Security Solutions

Taelor Sutherland is the Associate Editor at Security magazine. Sutherland covers news affecting enterprise security leaders, from industry events to physical & cybersecurity threats and mitigation tactics. She is also responsible for coordinating and publishing web exclusives, multimedia content, social media posts, and a number of eMagazine departments. Sutherland graduated in 2022 with a BA in English Literature from Agnes Scott College.

https://www.securitymagazine.com/articles/102449-product-spotlight-on-mobile-security-solutions

CareCloud Data Breach Impacts Over 350,000

Healthcare information technology company CareCloud is notifying at least 350,000 people that their information was stolen in a data breach.

The incident involved an electronic health record environment within the CareCloud Health division, which was disrupted on March 16, 2026.

CareCloud’s investigation determined that hackers accessed one of its AWS environments between March 10 and March 16, and likely exfiltrated data from it.

On June 24, the investigation determined that personal, financial, and medical information was compromised in the incident, the company notes in the notification letter sent to the potentially affected individuals, a copy of which was filed with the Massachusetts Office of Consumer Affairs and Business Regulation.

The potentially affected information, it says, includes names, addresses, Social Security numbers, dates of birth, driver’s license numbers, government ID numbers, financial account numbers, credit/debit card numbers, and medical and health insurance information.

For a limited number of individuals, the compromised data also included full credit card information, including CVV, the company notes in an incident notification. CareCloud says it has no evidence that the stolen data has been misused.

Advertisement. Scroll to continue reading.

Based on filings with the Attorney General’s Offices in several states, at least 350,000 individuals have had their information stolen.

The company is providing them with up to 24 months of free identity theft protection, credit monitoring, and ID theft recovery services, which include a $1,000,000 insurance reimbursement policy.

“CareCloud engaged external cybersecurity experts and, with their assistance, secured the affected environment, eliminated the threat, and confirmed that no persistent unauthorized access remained. CareCloud is continuing to strengthen the security of its systems and environments,” the company says.

The healthcare organization has yet to share the total number of impacted individuals and details on the threat actor responsible for the attack. SecurityWeek has emailed CareCloud for additional details and will update this article if the company responds.

*Updated with additional information and link to CareCloud’s incident notice.

https://www.securityweek.com/carecloud-data-breach-impacts-over-350000/

Critical Code Execution Vulnerability Patched in TeamCity

JetBrains this week rolled out patches for a critical-severity vulnerability in TeamCity On-Premises that can be exploited without authentication.

Tracked as CVE-2026-63077 (CVSS score of 9.8), the security defect can be exploited via HTTP/S to bypass authentication and achieve remote code execution (RCE).

“An unauthenticated attacker could exploit the vulnerability via the TeamCity agent polling protocol to bypass authentication checks and execute arbitrary operating system commands with the privileges of the TeamCity server process,” JetBrains explains in its advisory.

Depending on the available privileges, an attacker could access TeamCity data, configurations, and credentials, could tamper with the server state, and could potentially compromise build artifacts and downstream CI/CD pipelines.

According to JetBrains, the flaw affects all TeamCity On-Premises versions. The company has already rolled out mitigations for TeamCity Cloud instances and has no evidence that the bug has been exploited in the wild.

“A fix for this vulnerability has been introduced in versions 2025.11.7 and 2026.1.3. We have also released a security patch plugin for 2017.1+ so that customers who are unable to upgrade can still patch their environments,” JetBrains announced.

Advertisement. Scroll to continue reading.

Users are advised to download and install either the latest version of TeamCity or the security patch plugin as soon as possible (the plugin resolves only this CVE, the company notes).

JetBrains also recommends limiting access to internet-facing TeamCity servers, running all servers with the minimum required operating system privileges, and using VPN connections or implementing additional protections to prevent unauthorized access.

“TeamCity servers should also run on dedicated hosts separate from build agents,” the company notes.

https://www.securityweek.com/critical-code-execution-vulnerability-patched-in-teamcity/

Why Security Fundamentals Still Matter in the Age of AI

When reports surfaced that Anthropic’s AI model, Mythos, could identify thousands of previously unknown vulnerabilities across critical software systems, the response was immediate. Regulators mobilized. Financial leaders convened emergency briefings. Headlines warned of a new era of cyber risk.

The concern is understandable, and the capabilities are significant. But the reality is that Mythos did not introduce an entirely new category of risk. AI has already been reshaping the cybersecurity landscape for years, making attacks faster, more scalable, and harder to detect; compressing the window between vulnerability disclosure and active exploitation from years to hours in some cases.

AI Is Accelerating Existing Threats

AI-powered cyberattacks did not begin with Mythos. Earlier-generation AI systems had already demonstrated the ability to identify vulnerabilities, automate reconnaissance, and improve phishing and social engineering attacks at scale.

What is changing now is the timeline. Vulnerability management teams used to have days or weeks to assess and remediate flaws. In some cases, automated scanning tools, increasingly augmented by AI, are probing the internet for newly disclosed vulnerabilities within hours or days of public identification. Organizations are racing against a clock that resets at the moment a CVE (Common Vulnerabilities and Exposures) goes public.

At the same time, AI is lowering the barrier to entry for attackers. Bad actors no longer need the same level of coding expertise to identify vulnerabilities and weaponize publicly disclosed flaws.

For phishing and social engineering attacks specifically, AI can help attackers create emails and messages that sound natural, polished, and personalized to the recipient. Instead of relying on obvious, generic scams, cybercriminals can produce convincing messages that are significantly harder to detect.

AI Still Exploits the Same Weaknesses

Despite the headlines, AI-driven attacks continue to rely on familiar security gaps: weak or shared credentials, missing multi-factor authentication (MFA), poor network segmentation, excessive access permissions, and unrestricted external connectivity

The attack methods may evolve, but the entry points are the same. In many cases, straightforward controls significantly reduce exposure. IP allowlisting, for example, can prevent entire categories of automated scanning and exploitation tools from ever reaching critical systems. These controls are well established. They still stop attacks when they’re implemented and maintained properly.

Yet, organizations continue to be compromised because those fundamentals are never fully implemented. That failure carries a higher price than it once did. Gaps that organizations previously had time to absorb are now critical exposures. Automated tools are scanning for and targeting newly disclosed vulnerabilities within hours of disclosure. The margin for incomplete compliance cyber hygiene has effectively disappeared.

Compliance Is a Security Strategy: Why PCI Still Matters

PCI DSS is built around the security controls that attackers continue to target. Those controls work because attackers still depend on the same security gaps to gain access. Its 300+ technical controls are intentionally rigorous, designed to reduce the likelihood and impact of exactly the kinds of attacks AI is now making easier to launch at scale.

There is a consistent pattern across major breaches. More often, the required controls were incomplete, inconsistently enforced, or treated as a compliance exercise instead of a security strategy. Research has shown that fully PCI-compliant organizations are far less likely to experience payment card data breaches.

Controls like network segmentation with strong access control, firewall configuration with default deny all loosened only to business necessary traffic, EDR, and least privilege remain highly effective against AI-augmented attacks because good security architecture anticipates how attackers operate, regardless of the tools they use. AI may be effective at network exploitation, but it’s not necessarily quiet, so detection is important.

The industry’s instinct will be to respond to AI threats with more AI tools. In many cases, that instinct is justified. Attackers are already using AI to find vulnerabilities faster. Defenders should be using it for the same reason. That does not mean replacing security discipline with another tool. It means using AI to pressure-test systems, find exposed assets, prioritize remediation and shorten the time between discovery and response. But it only helps if the basics are already in place. Access controls, segmentation, monitoring, patching and compliance still determine whether an attack becomes a breach.

What Organizations Should Prioritize Now

To strengthen defenses against AI-driven threats, organizations should focus less on the headlines and more on operational discipline by taking the following steps:

Conduct an honest assessment of PCI DSS compliance: Treat compliance as a real evaluation of where controls may have drifted, weakened or were never fully implemented. Pay particular attention to access controls, privileged account management, and network segmentation.
Tighten your vulnerability management program for the new timeline: Quarterly patching cycles are no longer sufficient when attackers can now exploit newly disclosed vulnerabilities within hours. Know every system and device in your environment. Anything you can’t see may be a vulnerability that attackers find first. Organizations should also evaluate where AI can help reduce the time between discovering a vulnerability and remediation.
Focus on the security gaps attackers target most often: Eliminate shared credentials, enforce strong password policies, implement multi-factor authentication consistently, and restrict network access to what is explicitly authorized for business purposes.
Take AI-augmented phishing seriously: Social engineering attacks are becoming more targeted and harder to identify as fraudulent. Employee awareness training and verification protocols need to reflect that reality.
Maintain Human Oversight and Expertise: AI can surface vulnerabilities faster, but organizations still require skilled people to validate the risk and decide how to respond.

AI shortened the time security teams have to respond. It has not replaced the need for strong security controls. Organizations that consistently execute the fundamentals will be better prepared for what comes next.

Mythos did not rewrite the principles of payment security. It made the consequences of ignoring them, or weak execution, far more immediate.

https://www.securitymagazine.com/articles/102417-why-security-fundamentals-still-matter-in-the-age-of-ai

CISA Urges Water Sector to Protect OT After Coordinated Attacks on PLCs

The US Cybersecurity and Infrastructure Security Agency (CISA) is urging water and wastewater system (WWS) operators to protect operational technology (OT) against malicious activity targeting programmable logic controllers (PLCs).

The alert is a fresh call to action that comes just days after a coordinated cyberattack disrupted automated controls at dozens of water utilities in Minnesota.

In an alert published July 30, CISA said it is observing a significant increase in threat actors targeting PLCs in the water and wastewater sector, and urged critical infrastructure owners, operators, and integrators to remove publicly exposed PLCs and other OT from the internet as soon as possible.

The agency described specific tactics it has seen against exposed controllers: attackers have modified passwords to lock out operators and disconnected PLCs by changing their IP addresses. According to CISA, the attacks have resulted in “boil water notices” and sustained manual operations that closely mirror what several Minnesota utilities reported this week.

Notably, the alert stressed that the targeting spans water entities of all sizes, and that even organizations with mature cybersecurity programs should validate their external connections. CISA specifically called out cellular modems installed by operators, vendors, or system integrators that may not be documented or captured in routine attack surface scans.

Minnesota attacks underscore the warning

CISA’s renewed push follows a coordinated cyberattack that, according to Minnesota IT Services (MNIT), hit OT systems at more than 30 community water systems on July 26 and 27.

Advertisement. Scroll to continue reading.

As SecurityWeek reported earlier this week, statements from affected cities, including Maple Plain, Braham, South St. Paul, and Plymouth, indicated that some automated control functions were disrupted, though contingency procedures were activated and water and wastewater operations remained functional in most cases. The affected cities told residents that drinking water remained safe.

State and federal agencies are investigating, and no formal attribution has been made.

Ties to the Iranian PLC campaign

The timing of the Minnesota intrusions is notable. They came shortly after the US government warned critical infrastructure organizations about Iran-linked attacks on industrial control systems made by Siemens, Rockwell Automation, and Schneider Electric.

That warning came via a July 22 update to advisory AA26-097A, originally published in April, which expanded the list of targeted vendors beyond Rockwell Automation’s Allen-Bradley controllers to include Schneider Electric and Siemens devices, and noted that PLCs from other manufacturers may also be at risk. Investigators have observed activity against Rockwell CompactLogix and Micro850, Schneider Electric Modicon M340, and Siemens S7-1200 series PLCs.

Iranian threat groups including CyberAv3ngers and Handala fit the profile for attacks on water systems of the kind seen in Minnesota, though investigators have not linked the incidents to any specific actor. CyberAv3ngers has a long track record of targeting small water utilities and municipal facilities, and in 2020 attacks on water facilities in Israel, Iran-linked actors exploited vulnerable cellular routers as an entry point.

What OT operators should do

CISA’s core message to the sector is unchanged but increasingly urgent: internet-exposed OT must be secured. The July 30 alert recommends three immediate steps. Operators should disconnect the PLC from the internet, routing any remote access for operational purposes through a VPN or gateway device rather than directly to the controller; enable password protection and change default passwords; and allowlist IP addresses so that remote access is permitted only from known engineering laptops or other critical OT assets.

CISA also advised that, after disconnecting PLCs, operators ensure they have a known-clean backup of the PLC image in case they are locked out by a modified password. Owners and operators of Rockwell Automation MicroLogix 1400 controllers are pointed to Rockwell’s dedicated guidance for restoring access when the password is unknown.

Beyond the immediate steps, utilities are encouraged to review the tactics, techniques, and indicators of compromise in AA26-097A for signs of current or historical activity on their networks. For more information, read the full alert from CISA.

Learn More at the ICS Cybersecurity Conference | Nashville

https://www.securityweek.com/cisa-urges-water-sector-to-protect-ot-after-coordinated-attacks-on-plcs/

Bank of America to Acquire Cybersecurity Firm MDSec

Bank of America announced on Thursday that it plans to acquire UK-based information security consultancy MDSec Consulting Limited.

Headquartered in Macclesfield, England, MDSec provides technical information security consulting services and employs approximately 65 cybersecurity professionals.

The acquisition will expand Charlotte, North Carolina-based Bank of America’s existing presence in northern England. The financial institution has more than 1,400 employees in nearby Chester, where it also operates a cyber threat operations center.

“We have long admired the exceptional ability of the MDSec team and are delighted that Bank of America and its clients will now further benefit from their work,” said Bank of America chief information security officer Kris Fador.

MDSec co-founder Dominic Chell said joining Bank of America would allow the company to further its efforts to develop security capabilities and advance technical innovation.

“Joining one of the world’s leading financial institutions, one that reflects our culture of innovation and technical excellence, gives us an incredible opportunity to take that ambition to the next level,” Chell said.

Advertisement. Scroll to continue reading.

The transaction is expected to close during the fourth quarter of 2026, subject to regulatory approvals. Financial terms were not disclosed.

https://www.securityweek.com/bank-of-america-to-acquire-cybersecurity-firm-mdsec/