Browser Security in the Age of Mythos: How AI Is Driving a Surge in Browser Risk Exposure

Large language models trained on coding tasks have become remarkably adept at finding security issues. Effective enough that Anthropic made the decision to restrict access to its latest model, Mythos. But between software vendors that have access to Mythos and the advances made by other models, it was broadly expected that we would see a large number of security vulnerabilities being found in the near future.

That future arrived faster than most expected. The patch tsunami is here, and browsers are absorbing the worst of it. This was also expected, given that browsers are the most exposed attack surface and that browser vendors have been at the forefront of vulnerability research.

Vulnerability Discovery Is Exploding

In the span of just over a month, Google released a steady stream of Chrome updates fixing more than 800 security issues: 33 on June 16, 27 on June 11, 74 on June 8, 429 on June 2, 151 on May 27, 16 on May 19, 79 on May 12. For its part, Mozilla fixed 22 security issues found in Firefox in February and then fixed a whopping 271 security issues at the end of April. The issues in Firefox were found by successive versions of Anthropic models: Opus 4.6 in February, Mythos Preview in April. Google has not publicly confirmed exactly how the Chrome vulnerabilities were found, but it is almost certain that LLMs were involved.

What makes this moment feel so different is the tempo. Both the pace and the number of vulnerabilities given a critical severity rating is unprecedented. In the case of Chrome, this has resulted in 7 separate updates in 6 weeks, with each update fixing at least one critical issue, something that would previously rarely happen even twice in 6 weeks.

Beyond Discovery: Exploitation and Evasion

This influx of security fixes is creating risk for people sitting behind browsers because the same models driving discovery have become increasingly capable at the next step: exploitation. The conversation around Mythos and other frontier models has often been focused on their ability to find vulnerabilities, including a 27-year old bug in OpenBSD. While this is impressive, just as important is how much the coding models can help in writing exploits. Turning a vulnerability into an exploit that can reliably compromise a target is far from trivial.

Mostly, security researchers stop at proof-of-concepts that can crash the browser. This is enough to convince a software engineer that there is a memory safety issue that could lead to an exploit. This leaves significant work that used to require time and specialized skills to weaponize a vulnerability. LLMs accelerate this timeline and lower the barrier to entry. Tasks that used to require deep knowledge can now be assisted and accelerated by the same class of models that are discovering the flaw.

Faster exploitation is not the only way LLMs are shifting the balance. A recent report from the Google Threat Intelligence Group highlights that frontier models are also adept at defensive evasion. Attackers have now started using models to mutate their attack payloads. Security experts have long known that signature based approaches were relatively easy to defeat for a moderately skilled attacker, but this is one more area where the barrier to entry is being lowered.

Rethinking Browser Security for an AI-Driven Threat Landscape

This points to a new reality where there is no time to deploy patches and where band-aid solutions in the form of signature-based attack detections cannot be relied on. Security leaders need approaches that fundamentally address the risks inherent in browsing complex, untrusted websites.

For browser vendors, this for example means moving to memory safe language to implement browsers, work that is already under way. For everyone else, it means moving the browser attack surface off endpoints and continuing to focus on security fundamentals, such as following the principle of least privilege. This implication shifts away from relying on detection and response to reducing the impact of compromise when they inevitably occur.

Moving the attack surface off the endpoint means that when a user visits a malicious site, the attack executes in an environment that disappears when the session ends. Nothing reaches the endpoint. Nothing persists. The model found the vulnerability, the attacker wrote the exploit (possibly with the help of the model), and the exploit ran inside a container that no longer exists. In that model, the goal isn’t to eliminate every attack path, but to make successful attacks have far less of an impact.

https://www.securitymagazine.com/articles/102455-browser-security-in-the-age-of-mythos-how-ai-is-driving-a-surge-in-browser-risk-exposure

6 Crisis Response Best Practices (That Actually Hold Up When Things go Sideways)

@import url(‘https://fonts.googleapis.com/css2?family=Open+Sans:ital,wght@0,300..800;1,300..800&display=swap’); @import url(‘https://fonts.googleapis.com/css2?family=Open+Sans:ital,wght@0,300..800;1,300..800&family=Roboto:ital,wght@0,100..900;1,100..900&display=swap’); @import url(‘https://fonts.googleapis.com/css2?family=Handlee&display=swap’); h3 { font-weight: bold; font-size: 18px; color: #C72026; } figure { padding: 4px; margin: auto; } figcaption { color: #404144; Roboto”, sans-serif; font-size: 14px; padding: 10px; text-align: left; } .pqFull { display: block; border-width: 2px 0; border-style: solid; border-color: #404144; padding: 1.5em 0 0.5em; margin: 1.5em 0; position: relative; Handlee”, Arial, sans-serif; font-size: 20px; text-align: center; color: #0067A6;
}
.pqFull:before { content: ‘\275D’; position: absolute; top: -.10em; left: 50%; transform: translate(-50%, -50%); background: #fff; width: 4rem; height: 2rem; color: #666; text-align: center;
}
.pqFull:after { content: “\2013 \2003” attr(cite); display: block; text-align: center; font-size: 18px; color: #404144;
}

Crises don’t follow scripts. A protest can drag on for weeks. An active conflict can shift by the hour. A natural disaster cleans up in days. The incident type changes, but the pressure on security teams doesn’t. So the question isn’t do you have a playbook, it’s whether your team can execute when reality doesn’t match page one.

1. Your Playbook Is Not Your Crisis Capability

A lot of organizations mistake documentation for readiness. They have a playbook, a notification tree, and a slide that shows the crisis team structure, but real capability comes from knowing how the organization will make decisions under pressure. Who owns life safety, operations, and communications? Who has authority to act before the full leadership group is assembled? What can be decided at the site, regional, or enterprise level? What triggers escalation?

The structure has to match the way the business actually works. If every decision still has to climb a slow approval chain, the response model is mostly cosmetic. A playbook can guide the response, but it cannot replace judgment, authority, or structure.

Most crisis exercises are too clean. The scenario is clear, the timeline is neat, and everyone knows they are in an exercise. Real crises never unfold that way.

2. Practice the Messy Middle, Not the Perfect Scenario

Most crisis exercises are too clean. The scenario is clear, the timeline is neat, and everyone knows they are in an exercise. Real crises never unfold that way. Information comes in fragments, facts change, stakeholders disagree, and teams have to make decisions before everything is confirmed. Training should prepare people for that part. Build exercises around incomplete information, unavailable decision-makers, competing priorities, bad assumptions, and communication gaps.

When I was a regional crisis manager managing 14 CMTs, the value of training was not checking the annual box. If someone didn’t show up to the tabletop, someone else had to fill that role, even if it was not their specialty. That is real life. Not everyone will be available when the incident happens. The exercise showed where the response slowed down, where ownership was unclear, and where people waited for permission instead of acting. A good exercise should make the real incident less surprising.

3. The Best Crisis Managers Have a Comms Bestie

Communication is one of the most critical parts of crisis management, and the relationship with your communications partner cannot start once the incident is already moving. As a crisis manager, your communications lead should be one of the closest relationships you maintain. Build trust before you need to move fast together.

In a crisis, communications becomes a pressure release valve for the organization. They help turn incomplete facts into clear, calm, usable information for leaders, employees, and stakeholders. The mistake is bringing them in only when a message needs approval. They need to be involved early, helping shape what is said, when it is said, who needs to hear it, and what tone the organization should take. Strong communication buys time, reduces confusion, and keeps trust intact while the response continues.

When the crisis ends, the work is not done. The teams that actually improve after a crisis are the ones that treat the post-incident review as non-negotiable.

4. Make Communication Two Way

Sending updates isn’t enough. You need to know who’s safe, who’s impacted, and who needs help. Build a mechanism for people to respond, such as a poll, a reply function, a simple check-in prompt. The information coming back from employees, travelers, site leaders, or field teams can quickly become one of the most valuable parts of the response. It helps you separate assumptions from reality, prioritize resources, and identify problems that may not show up in a dashboard. In a fast-moving crisis, the signal coming back from your people is as valuable as anything you’re pushing out.

5. Do Not Let the Lessons Learned Die in a Folder

When the crisis ends, the work is not done. The teams that actually improve after a crisis are the ones that treat the post-incident review as non-negotiable.

Survey the stakeholders who were involved. Ask what worked, what slowed the response, what was unclear, and what decisions were harder than they should have been. Present the findings. Assign owners. Track the action items. With daily work and new crises constantly adding to the workload, post-incident reviews can easily slip into the abyss. That is exactly why they matter. This is where the real wins are made.

CMTs should use this process to surface the things that get missed in the middle of response: outdated contact lists, unclear role ownership, communication breakdowns, approval delays, and assumptions that did not hold up. The point is not to blame people. The point is to make the next response tighter. And in crisis management, there is always a next time.

https://www.securitymagazine.com/articles/102457-6-crisis-response-best-practices-that-actually-hold-up-when-things-go-sideways

When Cyberattacks Hit Medical Devices, Patients Pay the Price

Healthcare organizations have spent years discussing cybersecurity through the lens of data breaches, ransomware payments, regulatory exposure, and reputational damage. Those risks matter. But when cybersecurity conversations begin and end with protecting data, we miss the bigger issue. In healthcare, the most serious consequence of a cyberattack is what happens to patient care.

As an emergency physician and Chief Medical Officer, I’ve spent my career working in environments where seconds matter and where clinicians depend on technology to make critical decisions. When systems become unavailable, unreliable, or difficult to access, care delivery changes immediately. Clinicians lose visibility into the information they need. Workflows break down. Communication becomes harder. Treatment can be delayed.

That reality becomes even more important as medical devices become increasingly connected to the broader healthcare ecosystem. Today’s devices are woven into the workflows clinicians rely on to diagnose, monitor and treat patients. When those devices are disrupted by a cyberattack, the consequences extend far beyond the technology itself.

That’s why medical device security should not be viewed primarily as an IT issue. It is a patient safety issue.

Medical Device Threats Are No Longer a Niche Threat

Recent research found that nearly one in four healthcare organizations experienced cyberattacks impacting medical devices over the past year. Of those incidents, 80 percent directly affected patient care. When medical devices are compromised, care must change in real time, often to its detriment. Safeguarding against such attacks is just as much about protecting endpoints or reducing organizational risk exposure as it is about keeping care delivery running safely. The issue with responding to such alarming statistics from an IT and compliance silo is that the two most common ways to address them (locking things down so much that they are unusable or working around them) can inadvertently cause the same damage to care delivery as the attacks themselves. The real solution needs to be good technology that secures the endpoints but remains easy to use.

I have seen firsthand how fragile clinical workflows can become when technology suddenly becomes difficult to access, is unavailable, or is untrustworthy in performing its intended function. Most people outside healthcare lack insight into how interconnected modern care delivery has become. Medical devices are deeply integrated into clinical workflows, authentication systems, electronic health records, medication administration processes, monitoring platforms, and communication systems.

What Care Disruption Actually Looks Like on the Frontlines

One of the realities of healthcare is that clinicians will always find a way to care for patients.

When technology becomes difficult to access, clinicians do not simply stop working. They adapt. They create workarounds. They share information through alternate channels. They delay documentation. They revert to manual processes. They do whatever they believe is necessary to keep care moving. That resilience is one of healthcare’s greatest strengths, but it can also create risk.

I have seen this firsthand throughout my career. In emergency medicine, clinicians simply cannot accept unnecessary delays when caring for critical patients. When devices become inaccessible, clinicians may delay documentation, share credentials, bypass standard authentication steps, revert to paper processes, manually transcribe information, or postpone diagnostics or procedures until systems become available again. Those decisions are made with the best intentions: keeping patient care moving under difficult circumstances. But every workaround introduces new opportunities for error, gaps in accountability, and additional security risk.

Consider what happens if an infusion pump network becomes unavailable because of a cyber incident. Medications may need to be administered through alternate workflows. Verification steps that are normally automated may become manual. Nurses may spend additional time documenting, double-checking information, or reconciling records across systems. This may sound simple. But, in reality, this is a huge burden. Healthcare environments already operate under significant cognitive and operational strain. Adding even small amounts of friction during a crisis increases the likelihood of mistakes and creates opportunities for delays, miscommunication, or error.

Security Must Work the Way Care is Delivered

Historically, healthcare cybersecurity strategies have focused heavily on protecting devices by patching endpoints, segmenting networks, and securing hardware. Those controls remain essential. But they do not fully address how care is delivered day by day, hour by hour.

A secure device that clinicians cannot access efficiently during patient care is still a problem. In fact, if security controls create enough friction that clinicians begin bypassing them, organizations can inadvertently increase risk rather than reduce it.

This is why cybersecurity decisions cannot be made in isolation from clinical operations. Every additional step, every delay, and every obstacle introduced into a workflow has the potential to affect patient care. Security controls that sound effective in theory can create unintended consequences at the bedside if they fail to account for how care is actually delivered.

Healthcare security cannot succeed if it ignores the realities of clinical workflows. That is where many organizations need to evolve their approach. They need to move beyond thinking only about devices and networks and focus equally on how clinicians access systems, move through workflows, and continue delivering care under pressure — without introducing unnecessary friction.

Cyber Resilience Is Clinical Resilience

Attackers understand that hospitals are uniquely vulnerable to operational disruption because patient care cannot simply stop. The need to get systems back online is far more critical by nature than in most other industries, making healthcare uniquely vulnerable to attacks designed to create urgency and chaos. Because many connected devices operate on legacy systems, are difficult to patch, or rely on complex vendor ecosystems that complicate visibility and accountability, medical devices expand the attack surface. The challenge is not only preventing these disruptions, but ensuring organizations can continue delivering safe care when they occur.

Hospitals respond more effectively during incidents when cybersecurity teams, clinical leadership, biomedical engineering, and frontline staff already know how to work together. Most organizations now conduct downtime drills for electronic health record outages, but far fewer meaningfully simulate medical device disruptions across interconnected workflows.

That gap means that there is rarely time to build safe processes in the middle of an incident. Organizations that regularly test response plans are better positioned to identify blind spots, clarify responsibilities, and understand how care delivery will continue when critical systems become unavailable. In many ways, resilience is built long before an attack occurs.

Healthcare organizations need to evolve their thinking. Medical device security matters because these technologies have become deeply embedded in how clinicians diagnose, monitor, and treat patients. Protecting them requires more than securing hardware or responding to incidents after the fact. It requires designing security, access, and operational processes that support the realities of care delivery, especially during periods of disruption.

Cyber resilience and clinical resilience are increasingly the same thing.

The organizations that will be best prepared are those that recognize cybersecurity as an integral part of patient safety strategy, not a separate technical function operating alongside it.

Because when a cyberattack affects a medical device, the most important question isn’t what happened to the technology. It’s what happened to the patient.

https://www.securitymagazine.com/articles/102453-when-cyberattacks-hit-medical-devices-patients-pay-the-price

Balance Theory Raises $19 Million to Help Enterprises Manage Cybersecurity Investments

Cybersecurity investment management startup Balance Theory has raised $19 million in Series A funding to expand its platform for helping CISOs evaluate and manage security spending.

Balance Theory’s platform is designed to bring cybersecurity investment planning, market intelligence and execution into a single system. It maintains contextual information about an organization’s security program, supplements it with proprietary market data, and employs AI agents and automated workflows to support purchasing and portfolio-management decisions.

As the company explains, “Balance theory manages security investment events end-to-end. Detect investment decision triggers, run each event for the best cost and coverage outcome, and continuously rationalize the program to maximize the impact of every dollar deployed.”

The platform also creates a record explaining why individual investments were made and monitors for changes that could affect their value, suitability or priority. According to the company, its technology currently manages more than $1 billion in cybersecurity spending.

The funding round was led by SYN Ventures, with participation from existing investors DataTribe and TEDCO. The company also announced that Dan Burns, founder of Accuvant and former CEO of Optiv, has joined as executive chairman.

Balance Theory said it will use the new funding to accelerate go-to-market efforts, build deeper enterprise integrations, expand its cybersecurity market intelligence, and continue developing the platform’s AI agents and skills.

Advertisement. Scroll to continue reading.

The Columbia, Maryland-based company raised $3 million in Seed funding back in 2022.

“Security leaders lacked a consistent way to understand their own enterprise, navigate an increasingly complex market and connect those insights to action,” Balance Theory co-founder and CEO Greg Baker said.

The company did not disclose its valuation following the investment.

https://www.securityweek.com/balance-theory-raises-19-million-to-help-enterprises-manage-cybersecurity-investments/

Ruby on Rails Patches Critical Vulnerability

Ruby on Rails this week rolled out patches for a critical vulnerability that could allow unauthenticated attackers to achieve remote code execution (RCE).

A server-side web application framework written in Ruby, Ruby on Rails is used for the fast building of full-stack web applications and APIs.

Tracked as CVE-2026-66066 (CVSS score of 9.5), the critical security defect is described as an arbitrary file read that potentially exposes secrets, allowing remote attackers to execute code or move laterally to other systems.

“In its default configuration, a Rails application that displays image variants may allow an unauthenticated attacker to read arbitrary files from the server, including the process environment,” Ruby on Rails’ maintainers note in an advisory.

Within the exposed environment, the advisory explains, attackers could find secret_key_base and credentials for external systems, which can be abused to escalate the attack to RCE.

The issue impacts applications that use the libvips library for Active Storage image processing and that allow image uploads from untrusted users.

Advertisement. Scroll to continue reading.

Because libvips marks some file read and write operations as ‘unfuzzed’ (unsafe for untrusted content) and Active Storage did not disable the unfuzzed operations, an attacker could upload a crafted file to invoke one of these operations.

“We are aware of a mechanism by which an attacker, by uploading a crafted file, is able to cause disclosure of the contents of arbitrary files accessible on the filesystem of the targeted application,” Ruby on Rails’ advisory reads.

CVE-2026-66066 was patched in Active Storage versions 7.2.3.2, 8.0.5.1, and 8.1.3.1. Users are advised to update their deployments as soon as possible, as well as to update libvips to at least version 8.13, as previous library releases do not support disabling unfuzzed operations.

“Upgrading closes the vulnerability but does not undo an exfiltrated secret if that already occurred. An affected application should treat every secret readable by the application process as potentially exposed and change it,” Ruby on Rails notes.

According to cybersecurity firm Rapid7, as of July 30, there is no evidence that the security defect has been exploited in the wild.

https://www.securityweek.com/ruby-on-rails-patches-critical-vulnerability/

The Execution Gap: Why Great Security Design Doesn’t Always Deliver Great Security

.pullquote{ width: 250px; padding: 20px;
} .pullquote a,
.pullquote h4 a:link,
.pullquote h4 a:visited { text-decoration: none; border-bottom: none;
} .pullquote h4 a:hover,
.pullquote h4 a:focus { text-decoration: none; border-bottom: 1px solid #000;
}
.pullquote h4{
}
.pqfloatleft{ float: left; margin: 0 15px 10px 0; border-right: 4px solid #d91920; text-align: right;
}
.pqfloatright{ float: right; margin: 0 0 10px 15px; border-left: 4px solid #d91920;
}
@media only screen and (max-width: 600px) { .pullquote{ margin: 20px auto 20px auto; width: 250px; display: block; } .pqfloatleft{ float: none; border-right: none; border-top: 2px solid #d91920; border-bottom: 2px solid #d91920; text-align: center;
}
.pqfloatright{ float: none; border-left: none; border-top: 2px solid #d91920; border-bottom: 2px solid #d91920; text-align: center;
}
}
.cap { font-size: .85em; text-align: center; margin-top: 0px; margin-bottom: 35px; color: #595959;
}
.line{ background-color: #d91218; height: 5px; margin-top: 5px;
}
.side ul{ padding-left: 20px; font-size: .85em; font-style: italic; margin: 10px 15px 10px 15px;
} .side h3{ background-color: #135783; padding: 15px; color: #eeeeee; text-align: center;
}
.side h4{ color: #d91218; text-align: center; margin-top: 20px;
}
.side { color: #595959; width: 300px; display: inline-block;
}
.side p { font-size: .85em; caption-side: bottom; padding: 10px 5px 0px 5px; }
.floatleft{ float: left; margin: 0 15px 0px 0;
}
.floatright{ float: right; margin: 0 0 0px 15px;
} @media only screen and (max-width: 600px) { .side{ margin: 20px auto 20px auto; width: 100%; display: block; } .floatleft{ float: none;
}
.floatright{ float: none;
}
}
.sidebyside p { font-size: .85em; caption-side: bottom; padding: 10px 5px 0px 5px; color: #808080; }
.sidebyside{ width: 50%; float: left; display: block; display: inline-block; margin-bottom: 20px; padding-right: 5px;
}
.clearfloat{ clear: both;
} @media only screen and (max-width: 600px) { .sidebyside{ width: 100%; float: none; display: block; }
} .bigside{ background-color: #E8E8E8; padding: 20px; margin: 20px auto 20px auto; border: 1px solid silver; border-radius: 10px;
}

Walk through almost any modern data centre and you’ll find some of the most advanced physical security technology ever deployed. Intelligent video analytics, integrated access control, sophisticated perimeter protection, and increasingly AI-enabled operations have transformed what our industry is capable of delivering.

At the same time, the standards governing these environments have never been stronger as most enterprise and hyperscale organisations have invested significant time defining how their facilities should be designed, installed, commissioned, and maintained. As a result, they have established global governance frameworks, comprehensive commissioning procedures, and rigorous quality checks in the process. In many respects, we know what good security should look like.

Yet despite this maturity, one theme continues to appear on projects around the world: the completed facility doesn’t always reflect the original design intent.

This means that we are seeing a fundamental shift in how we view data centre security. The question is no longer “How do we best secure our data centres?” Instead, it is “How do we consistently deliver the same standard across dozens of facilities, multiple continents, and increasingly compressed construction programmes?”

And that is a very different challenge.

After visiting data centres across North America, EMEA, Asia and Latin America, one observation continues to stand out. The standards are often remarkably aligned, but the execution rarely is. I don’t believe this is a technology problem, nor do I believe it is a standards problem. Instead, I believe it to be an execution problem, and as our industry continues to scale at unprecedented speed, execution has become one of the defining challenges facing global physical security programmes today.

Why Hyperscale Changes Everything

Demand for digital infrastructure continues to accelerate, driven by cloud computing, AI, and society’s ever-growing dependence on data. New campuses are being developed at extraordinary speed, often with multiple facilities under construction simultaneously across different regions.

The expectation from clients, however, remains unchanged. Whether a facility is delivered in Virginia, Dublin, São Paulo, Hyderabad, or Singapore, it should provide the same operational experience, the same resilience, and the same level of security.

Achieving that level of consistency is significantly more difficult than writing a global standard as every project introduces different contractors, local regulations, supply chains, labour markets, and construction practices. Standards may be global, but delivery is always local, and so challenge becomes ensuring those standards survive the realities of design, procurement, construction, commissioning, and handover.

And that is where many global programmes begin to diverge. Not because the design was flawed, but because consistently translating design intent into operational reality becomes increasingly difficult as speed, scale, and complexity grow.

The Execution Gap

One of the biggest misconceptions within our industry is that security issues begin during installation. In reality, they usually begin much earlier.

They begin when security is engaged after key design decisions have already been made, when ownership between construction trades is unclear, when communications infrastructure, electrical systems, and security systems are designed independently rather than collaboratively, and when procurement decisions are made without fully considering long-term requirements.

Individually, these decisions appear insignificant. Collectively, however, they determine whether the completed facility reflects the original design intent.

I believe many projects experience what I describe as “design drift,” or the gradual movement away from the original design intent through hundreds of well-intentioned decisions made during procurement, construction, commissioning, and handover. It is rarely the result of one major mistake, but more often the accumulation of many small decisions, each logical in isolation, but collectively moving the finished solution further away from the original vision.

Projects rarely drift because of one poor decision. They drift because of hundreds of reasonable ones.

Consider something as simple as an access-controlled door. A global client may specify a mortise stile lock with concealed cabling routed internally through the door leaf, creating a clean, maintainable installation that supports both security and long-term serviceability.

Visit another facility built to the same global standard and you may find the exact same functional requirement but delivered in a very different way. The lock operates correctly, access control functions as intended, but the control cable is routed externally using flexible armoured conduit across the hinge side of the door.

Neither installation prevents the system from working. Both may comply with local regulations and achieve the same immediate objective, but they are no longer the same solution.

Somewhere between the original design and the completed installation, a series of well-intentioned local decisions changed the outcome.

This principle extends far beyond a single door. It can be seen in cable containment, door hardware selection, life safety interfaces, commissioning methods, equipment positioning, and trade coordination. None of these decisions, in isolation, is likely to determine whether a project succeeds or fails. Together, however, they create variation that has a lasting impact on the larger security program.

Multiply those decisions across hundreds of doors, multiple buildings, and several countries, and standardisation gradually disappears. The issue isn’t that people are making poor decisions; it’s that different teams are making different decisions. And that distinction matters.

Because as inconsistency increases across a portfolio, so does complexity. Maintenance teams encounter different hardware configurations, commissioning documentation becomes less reliable, spare parts inventories expand, training requirements increase, and troubleshooting becomes more time-consuming. What began as a series of sensible local decisions gradually creates a programme that is more difficult and more expensive to operate.

More importantly, inconsistency introduces uncertainty. When every facility is delivered slightly differently, organisations can no longer assume that maintenance procedures, operational workflows, or future upgrades will be executed the same way across their portfolio. At hyperscale, where programmes span multiple countries and facilities, it quickly becomes a business risk rather than simply a technical issue.

Governance Is About Protecting Good Decisions

Governance often receives an unfair reputation, associated with additional meetings, more approvals, and increased documentation.

In my experience, however, good governance doesn’t slow projects down, it reduces future rework at the ground level.

Its purpose is to protect good decisions before they become expensive corrections. It establishes clear ownership early in the project lifecycle, identifies trade dependencies before installation begins, and ensures regional delivery teams understand not only what the standard is, but why it exists.

For organisations delivering programmes across multiple regions, governance provides something even more valuable: repeatability. And with this level of repeatability, it creates the discipline needed to keep regional delivery aligned with global standards.

This only becomes more important as our industry continues to evolve. AI and analytics will become more intelligent, automation will improve overall efficiency, and the integration of physical and digital security will continue to strengthen.

And while these developments are exciting, they don’t remove the need for consistent execution. If anything, increasing system complexity makes it even more critical because every new integration, automation, and intelligent workflow introduces additional dependencies that must be coordinated throughout design, construction, commissioning, and operations.

Technology alone cannot compensate for poor coordination. AI, for example, cannot resolve unclear ownership, and automation cannot replace effective programme management. The more sophisticated our technology becomes, the greater the need for disciplined execution.

The Competitive Advantage We Don’t Talk About Enough

The security industry rightly celebrates innovation. Manufacturers continue to develop remarkable products, software capabilities evolve rapidly, and new technologies continue to attract significant attention.

And while innovation will undoubtedly continue, I believe the next competitive advantage will come from organisations that are able to preserve design intent throughout every stage of delivery.

The value of that capability extends well beyond delivering technology that simply functions. It gives clients confidence that the intended design will be accurately reflected in the finished facility, regardless of where it is built or who delivers it. As a result, it reduces operational complexity, simplifies future maintenance and expansion, improves the predictability of lifecycle costs, and ultimately lowers long-term programme risk.

Perhaps most importantly, though, it creates trust. Trust that every new facility will meet the same standard, deliver the same experience, and perform as intended without requiring each project to reinvent established practices.

The companies that succeed over the next decade will not necessarily be those with the newest technology, but those capable of delivering the same quality. Because the true measure of a global security programme isn’t whether it can produce an excellent design one time, but rather whether it can be reproduced with the same outcome, regardless of geography, contractor, or programme scale.

Execution Will Define the Next Decade

I’ve been fortunate throughout my career to visit data centres across North America, Europe, Asia, and Latin America. Every project teaches something, every region has exceptional engineers, designers, and delivery teams, but every country approaches projects in slightly different ways.

That diversity is one of our industry’s greatest strengths. Delivering it consistently is one of it greatest challenges.

Technology will continue to evolve, and standards will continue to improve. And with that evolution presents one of our industry’s greatest opportunities to protect design intent throughout the entire project lifecycle. Because security is rarely judged by how well it was designed but rather by how well it performs once construction is complete.

At the end of the day, technology creates capability, standards define intent, but execution is what transforms vision into operational reality.

Terry Browne

By Terry Browne, General Manager – Global Data Centre Group, Northland Controls

https://www.securitymagazine.com/articles/102477-the-execution-gap-why-great-security-design-doesnt-always-deliver-great-security

AI in Healthcare: Reducing Risk in the Emerging Malpractice Frontier

Artificial intelligence is transforming healthcare. From diagnostic imaging to clinical decision support, generative AI tools are increasingly embedded in patient care workflows. The benefits are significant: faster diagnoses, reduced physician burnout, and more personalized treatment. But with these advances comes a new category of legal risk: AI-related malpractice liability. Organizations that fail to approach AI adoption with the same rigor they apply to other clinical tools may face negligence claims that existing risk frameworks were never designed to address.

How AI Is Reshaping Malpractice and Negligence Claims

Traditional medical malpractice law rests on a well-established framework: a provider owes a duty of care to the patient, and liability attaches when the provider’s conduct falls below the applicable standard of care, causing injury. That standard has been measured against the knowledge, skill, and judgment of a reasonably competent practitioner in the same specialty.

Generative AI complicates this framework. In the event of an adverse outcome, the question of who bears responsibility—provider, health system, AI tool developer, or product manufacturer — becomes genuinely difficult. Time will tell how courts will allocate liability, but plaintiffs’ attorneys are exploring theories grounded in negligent adoption, inadequate oversight, and failure to verify AI outputs.

Critically, providers cannot delegate clinical judgment to an algorithm and expect it to absorb liability. AI does not hold professional licensure and is not subject to ethical standards that bind licensed practitioners. State licensing and scope-of-practice rules increasingly require licensed professionals to review and approve outputs, and expanding healthcare AI laws reinforce that AI cannot be the sole basis for clinical decisions. Providers who over-rely on outputs without applying their own expertise may face claims that they abdicated professional responsibility.

Beyond malpractice, health systems and AI developers face products liability exposure. Traditional products liability distinguishes between a product, its user, and the patient. AI blurs those boundaries; the chain of liability runs from hardware to software to developer to manufacturer to human provider, making fault allocation far more complex than in conventional medical device cases. Strict liability theories — including manufacturing or design defect, failure to warn, negligence, and breach of warranty — may all be viable. An open question is whether the standard of care will be higher for AI-enabled products because the software is arguably more “intelligent” than a reasonably prudent person, potentially raising expectations about what constitutes a defective or substandard output or a “reasonable” clinical provider.

Shadow AI

Perhaps the most underappreciated risk is the proliferation of unapproved AI tools. Staff may turn to publicly available AI platforms to summarize patient records, draft clinical notes, or research treatment options. These tools are often adopted without institutional knowledge or authorization and lack safeguards like encryption and closed-loop data handling.

Shadow AI creates legal and compliance exposure on multiple fronts. Unapproved tools have not been vetted for clinical accuracy and may violate organizational policies, regulatory requirements, and contractual obligations. If an adverse outcome is linked to reliance on an unapproved tool, the organization may face liability for both the clinical error and its failure to establish adequate AI governance.

Enforcement actions have also targeted process failures — including inadequate vendor diligence and failures of transparency — not just adverse outcomes. State attorneys general are not waiting for demonstrable patient harm before investigating alleged false and misleading product claims.

Security and Privacy Risks

There is a fundamental tension between familiar privacy principles — minimum necessary and data minimization — and AI’s operational demands for data retention to support explainability, bias analysis, and transparency. Generative AI tools require large volumes of data to function effectively. Organizations must evaluate whether AI tools transmit patient data to external servers, data is used to train third-party models, or adequate encryption and access controls are in place.

Under HIPAA and state privacy laws, organizations that fail to safeguard personal information processed by AI systems face significant penalties, potential algorithmic disgorgement, and litigation. A pivot to deidentified data does not solve these risks. The process of deidentifying data is itself a “use” of data, and certain state laws require consumer consent before deidentifying data and specific contractual flow-down terms with recipients.

Practical Steps

Providers do not need to avoid AI to manage these risks, but they must be deliberate. Organizations can take several steps to reduce liability exposure:

Establish formal AI governance committees that include clinical, legal, compliance, and information security leadership. AI tools should not be deployed without institutional review and approval.
Develop standards addressing patient notification, clinician obligations to independently verify AI-generated outputs, and permissible use of AI-enabled products, supported by training.
Conduct due diligence on AI vendors: understand how models are trained, what data rights vendors retain, how outputs are validated, and what ongoing monitoring the vendor supports. Use the HHS AI transparency rule “nutrition label” questions as a guide and update commercial contract terms, BAAs, and DPAs to address AI-specific concerns.
Implement ongoing monitoring protocols. AI systems change over time. Monitor for output drift, bias emergence, data quality degradation, and scope creep. Documentation creates critical evidence of diligence in any future enforcement action or litigation.
Treat AI incidents like safety events, not merely IT glitches, as these incidents can be a catalyst for class action litigation, regulatory enforcement, and lasting reputational harm.

The integration of AI into healthcare is not a question of whether but how. Organizations that invest now in governance, training, and oversight will be better positioned to capture AI’s benefits while managing the accompanying risks.

https://www.securitymagazine.com/articles/102420-ai-in-healthcare-reducing-risk-in-the-emerging-malpractice-frontier

How Election Manipulation Is Shifting from Hacks to Headlines

Modern society has collapsed the way it consumes news, education and entertainment into a single feed. Devices have surpassed television and radio as the most common source of news. More than half of U.S. adults say they get news from social media, and 86% access it from smartphones, computers, or tablets, giving platforms without traditional gatekeepers mass reach. The low barriers to publishing in these environments mean journalism, commentary, satire, activism and manipulation all compete side by side for attention.

This consolidation of content has redefined how misinformation can spread. One of the clearest reflections of this shift occurred during the 2016 election when a Russian-led campaign used high scale social engineering attacks, a cybersecurity threat that targets human behavior rather than system vulnerabilities, to manipulate voters, journalists, donors, and campaign staff. It proved that just securing networks wasn’t enough.

As a result, election security has shifted from protecting systems to protecting trust. Social media platforms and content providers, like campaigns or advocacy groups, are now at the center of how information is distributed, interpreted, and believed across audiences.

This dynamic is intensifying as the next election cycle begins, with the primaries starting now and the midterms in November. The digital environment has become a prime target for interference, with attackers weaponizing AI to scale their influence. Defending against these threats requires more than traditional security measures. It calls for enhanced practices from platform and content provider leaders, as well as greater awareness from individual voters.

From Hacking Systems to Shaping Perception

The 2016 election illuminated how the downstream effects of compromise can be as damaging as the initial breach, if not more so. It marked a turning point in election security, combining the theft of information with the planting of information.

Russian actors hacked political organizations and released real documents, while coordinated personas amplified divisive narratives online. The leaks lent credibility to the messaging, and the messaging shaped how they were interpreted, blurring the line between espionage and influence. This revealed a new objective to shape attention, credibility, and decision-making, turning distribution and credibility into key targets for manipulation.

Accountability Shift: Who Controls What We Believe?

Social media platforms such as Meta, X, TikTok, or YouTube control distribution, amplification, and discovery. Content providers like campaigns, journalists, influencers, and advocacy groups supply narratives that shape perceptions. Together, they determine what information is seen, believed and acted on.

This interdependence creates a system that is both powerful and fragile. Attackers can impersonate creators, manipulate content, and exploit algorithms to amplify misleading narratives. For example, ahead of the 2024 election, a video that used AI voice cloning to mimic Vice President Kamala Harris circulated, using visuals similar to an official campaign ad but with fabricated statements. This was shared on social media without clear parody labeling, highlighting how easily synthetic media can deceive and reach large audiences before context catches up.

Each player in this dynamic shapes trust in different ways. Platform providers serve as gatekeepers of visibility, as their algorithms shape the information environment with great reach. Content providers’ audiences rely on their credibility, voice and authority to interpret that information. When either layer is compromised, the impact compounds. Attackers’ ability to now use AI generated impersonation tactics to distort narratives significantly heightens this risk. The result is a shared and escalating challenge for security leaders at both platform and content providers.

The Security Leader’s Playbook for Election Integrity

The ultimate goal of influence campaigns is to shape the beliefs and behavior of individual voters, making mass communication platforms their most effective means of delivery. Now, security leaders at platform and content provider organizations must step in, as they sit at the front lines where operations either stall or gain traction.

For platforms, the priority is speed and coordination. They must strengthen election-related takedown processes by improving detection models, accelerating escalation, clarifying policies tied to electoral timelines, and coordinating cross-platform responses to limit the spread and lifespan of deceptive content.

For content providers, the focus is on elevating authenticity from an assumption to a practice. By authenticating and archiving original content, maintaining authoritative sources, publishing verifiable provenance signals, and preparing rapid-response workflows, they can strengthen their social engineering defenses and more efficiently counter manipulated or misleading content. Election campaign teams can reinforce these efforts with resources like Defending Digital Campaigns, which supports election security by helping campaigns implement core cybersecurity practices, offering free tools, training, and practical guidance to defend against digital threats.

Influence campaigns can reach voters across any channel, so resilience must also extend beyond institutions. Individuals must strengthen their personal defenses against misinformation campaigns, and establishing simple habits can make a significant difference. For example, before reposting or sharing anything widely, treat highly emotional, urgent, or surprising content as “unverified” until it’s been certified by a trusted source, such as an official campaign channel or a reputable news organization. Additionally, prioritize building media literacy skills, like recognizing when a headline is designed to provoke an emotional response, seeking original sources instead of screenshots, and staying vigilant around content that appears only in fringe or closed communities. Over time, these practices create the necessary personal buffer that reduces the spread of misinformation voters receive.

The Real Battleground Is Trust

As security strategies evolve from protecting infrastructure to securing a fast-moving, highly concentrated information environment, trust has become the most critical layer to defend. Modern strategies must account for how attackers shape perception through social engineering, impersonation, misinformation, and AI-generated content.

Defending elections now requires protecting three layers: the systems that run them, the narratives people consume, and the people who decide what to believe and share. For security leaders overseeing communication platforms, this is now central to their role, as they operate where credibility either holds or breaks. How they manage that risk will determine resilience in high-impact and everyday moments.

https://www.securitymagazine.com/articles/102418-how-election-manipulation-is-shifting-from-hacks-to-headlines

In Other News: OpenAI Open Source Tool, AWS Links Hacks to North Korea, Mythos Crypto Research

SecurityWeek’s weekly cybersecurity news roundup offers a concise overview of important developments that may not receive full standalone coverage yet remain relevant to the broader threat landscape.

This curated summary highlights key stories across vulnerability disclosures, emerging attack methods, policy updates, industry reports, and other noteworthy events to help readers maintain a well-rounded awareness of the evolving cybersecurity environment.

Here are this week’s highlights:

OnTrac hacked

Parcel delivery company OnTrac is notifying customers after attackers accessed its corporate network and certain files between March 20 and 22. The firm detected the activity on March 23 and engaged a third-party specialist to investigate the scope. No ransomware group has claimed the incident.

Advertisement. Scroll to continue reading.

Adobe patches vulnerabilities in Bridge, Campaign Classic and Format Plugins

Adobe issued security updates addressing multiple critical vulnerabilities, including a heap-based buffer overflow in Format Plugins that enables arbitrary code execution, several flaws in Bridge allowing code execution and privilege escalation, and Campaign Classic flaws that permit arbitrary code execution and file system reads. The Campaign Classic patch carries Priority 1 rating for on-premise deployments. Adobe reports no known exploitation in the wild.

SonicWall VPN and firewall accounts hit by widespread credential stuffing

Huntress observed a broad credential stuffing campaign against SonicWall VPN and firewall accounts beginning July 25, with successful logins at 30 organizations so far. The activity originates from five DigitalOcean-hosted IP addresses and appears automated, with no post-compromise hands-on activity detected.

OpenAI releases open source Codex Security CLI

OpenAI has open-sourced the Codex Security CLI, a tool for scanning repositories, tracking findings across runs, verifying fixes, and integrating security checks into CI/CD pipelines. The early release is available via npm and GitHub, with the company inviting feedback as it continues development.

UK Department for Education loses 607,000 contact records

Hackers obtained approximately 607,000 records containing phone numbers and email addresses from the Department for Education in England. The department says the data does not include bank details or other sensitive information, the incident was contained quickly, and the risk to individuals is not considered high.

Amazon ties Axios, Debug and Chalk hacks to North Korea’s Sapphire Sleet

Amazon Threat Intelligence attributes the recent compromises of the popular Axios, Debug, and Chalk NPM packages, along with a typo-crypto incident, to the North Korean group tracked as Sapphire Sleet. AWS notes the group’s focus on high-download packages for broad downstream impact and highlights evolving supply-chain techniques including fragmented payloads and environment-aware malware.

Researcher seizes control of Volvo/Eicher vehicle management platform

A security researcher discovered unauthenticated internal APIs in VE Commercial Vehicles’ My Eicher platform that exposed customer, user, and vehicle data and enabled account takeover. VE Commercial Vehicles is a joint venture between Volvo Group and Eicher Motors. The flaws allowed full control over fleets of commercial vehicles in India and access to sensitive documents such as Aadhaar cards. The primary issues were fixed after disclosure, and the company later remediated additional concerns.

Claude Mythos uncovers stronger attacks on HAWK and reduced-round AES

Anthropic researchers using Claude Mythos Preview developed an improved key-recovery attack on the post-quantum signature scheme HAWK that roughly halves its effective security level, and a faster meet-in-the-middle attack on 7-round AES. Neither result affects currently deployed systems—HAWK is still a candidate and the AES work targets a reduced-round variant—but both demonstrate AI-assisted progress in cryptanalysis.

https://www.securityweek.com/in-other-news-openai-open-source-tool-aws-links-hacks-to-north-korea-mythos-crypto-research/

Cyberattacks on Minnesota Water Systems Investigated as Officials Warn About Iranian Hackers

Authorities were working Thursday to find the source of cyberattacks that targeted over 30 water systems in Minnesota and came amid warnings that Iranian hackers have been focused on such systems.

Minnesota IT Services said state officials had yet to identify who was behind the attacks that took place Sunday and Monday. There were no reports that residents had been impacted by the attacks, though one city asked residents to conserve water for a couple hours while they tried to determine what was wrong.

The FBI, which is investigating, has not publicly identified a culprit and a spokesperson declined to say Thursday who the bureau thought might be responsible. The FBI, Cybersecurity and Infrastructure Security Agency and other agencies warned in an advisory last week that Iranian hackers have been targeting water and wastewater systems and the operational controls other critical infrastructure sectors.

Digital warfare has become ingrained in military conflict, and local water plants or healthcare facilities often lack the funds and know-how to install the latest software patches or take other security steps. That has made them a favorite target, both because of the relative ease of penetrating them and because of the panic such disruptions can cause.

[ Read: CISA Urges Water Sector to Protect OT After Coordinated Attacks on PLCs ]

Iran has the “geopolitical motivations” and a recent history of targeting water systems, said Cynthia Kaiser, the former deputy assistant director of the FBI’s cyber division who has been closely monitoring threats to critical infrastructure from hackers linked to the country.

Advertisement. Scroll to continue reading.

“I think most credible researchers and responders would be right to treat it like it’s Iran until proven otherwise,” added Kaiser, who is now the senior vice president of Halcyon’s Ransomware Research Center. “When it walks like a duck and talks like a duck, it’s really important to call it out.”

Iran’s interest in the operations of water systems inside the U.S. dates back years. In 2016, the Justice Department charged a group of Iranian hackers in connection with a cyberattack targeting a small dam near New York City.

Minnesota IT Services said that as of Thursday, there were no active requests from Minnesota communities for residents to modify usage of their drinking water. The state agency said most of the confirmed attacks involved technology that water systems use to remotely monitor and control equipment. It said that being impacted meant investigators confirmed there was malicious activity involving the system’s technology and didn’t mean every impacted community had their water service disrupted.

The state agency said there are similarities among the incidents, including timing and the types of technology used, but investigators haven’t determined if the same culprit was behind each one.

For a few hours on Monday, the city of Braham asked residents to minimize water use as they tried to determine why the water plant was offline. The city of about 1,700 people, located about 70 miles (113 kilometers) north of Minneapolis, said in a news release that the water plant outage was due to a cyberattack, but it didn’t cause any issue with water quality.

The city said the attackers shut down the operating controls that shut down the well and water treatment plant. That left the city for a time only able to provide residents with the water held in the water tower.

In Plymouth, a city of about 80,000 located outside of Minneapolis, officials said on social media that their water infrastructure communications had been restored by Tuesday afternoon following a cyberattack. The city said crews were able to continue operating the system during the outage and it didn’t have any impact on water levels or quality.

https://www.securityweek.com/cyberattacks-on-minnesota-water-systems-investigated-as-officials-warn-about-iranian-hackers/