OpenAI’s New Cyber Model and the Daybreak Split

OpenAI has unveiled GPT-5.6 Cyber, a large language model purpose-built for cybersecurity work, and it’s not just a new model—it’s a whole new access philosophy. The company is restructuring its Daybreak program into two tiers: Daybreak Blue for defensive tasks, and Daybreak Red for more advanced defensive work plus offensive cyber operations.

The move, announced on August 10, signals a deliberate attempt to balance capability with caution. It’s a recognition that the same AI power that can patch a vulnerability can also exploit one.

Daybreak Blue members get access to frontier general-purpose models like OpenAI‘s GPT-5.6 Sol, but with what the company calls “system-level safeguards” removed. That access is meant for authorized defensive work: vulnerability discovery, secure code review, malware analysis, incident response, and patch validation.

Daybreak Red, on the other hand, unlocks purpose-trained cyber models—including GPT-5.5 Cyber and the new GPT-5.6 Cyber—for heavier lifting like vulnerability research, exploit validation, and security testing.

It’s a two-lane highway, and OpenAI is betting that separating the traffic reduces the chances of a head-on collision.

What Makes GPT-5.6 Cyber Different

According to OpenAI, GPT-5.6 Cyber is a step-change from its predecessors. In internal evaluations, it completed 95% of a set of sensitive requests—covering exploit-chain development, authentication bypass, privilege escalation, and other advanced scenarios.

Compare that to general-access GPT-5.6 Sol, which completed just 1.5% of the same tasks. Even with Daybreak Blue access, Sol only hit 2.0%. The previous cyber-focused model, GPT-5.5 Cyber, managed 57.3%.

OpenAI also benchmarked GPT-5.6 Cyber against its peers on cybersecurity-specific tests like ExploitGym and ExploitBench, plus tasks such as zero-day vulnerability discovery and vulnerability reporting. It outperformed every other OpenAI model across the board.

In a real-world demonstration, OpenAI used GPT-5.6 Cyber to identify CVE-2026-15903, a high-severity vulnerability in V8, Chrome’s JavaScript engine. The company says it validated the finding and reported it to Google, which fixed the issue.

The Guardrail Paradox

Here’s the tension: OpenAI has deployed system-level safeguards on GPT-5.6 Sol to screen cybersecurity-related requests and prevent misuse. But those same guardrails can block legitimate defensive work.

Daybreak Blue access removes those guardrails for authorized users. Yet even without them, GPT-5.6 Sol will still refuse some “highly dual-use” prompts—like pentesting production systems. That’s where GPT-5.6 Cyber, exclusive to Daybreak Red, steps in.

The model is designed to reduce refusals on sensitive but legitimate tasks. It’s a fine line, and OpenAI is walking it carefully.

Industry Reaction: A Step Forward, Not a Solution

Alex Goller, principal solution architect for EMEA at Illumio, called the two-tier system “a good first step for OpenAI to mitigate two issues.”

“Daybreak Red restricts models from doing more harm than they should and limits people’s exposure to knowing what the models do,” he said. “Daybreak Blue fixes the gap that left Hugging Face’s responders unable to use frontier models during their incident.”

But Goller was quick to note that guardrails on a model are not the same as controls in your environment. “These are agents operating inside your environment and the controls that matter follow zero trust principles,” he explained. “There must be visibility into what the agent is doing and what it can reach, and then segmentation to contain the blast radius when something goes wrong.”

He praised OpenAI’s own guidance on sandboxing and scoped authorization, but added: “That enforcement lives in your infrastructure, not in the model.”

In other words, the model might be smarter, but your network still needs its own walls.

What This Means for Security Teams

For defenders, the takeaway is practical. If you’re in Daybreak Blue, you can now use frontier models for authorized defensive work without tripping over safety filters. That’s a real productivity win for incident response and code review.

For those in Daybreak Red, the offensive capabilities are more powerful—and more dangerous. Access is presumably restricted, but the power is there.

OpenAI’s move also highlights a broader trend: AI models are becoming specialized tools, not just general-purpose chatbots. The OpenAI cybersecurity model is a clear sign that the company sees security as a domain worth dedicated investment.

Still, the responsibility doesn’t end with the model. As Goller put it, the controls that matter “follow zero trust principles.” Visibility, segmentation, and containment are on you, not on OpenAI.

The Bottom Line

GPT-5.6 Cyber is a significant technical leap, and the two-tier Daybreak program is a thoughtful attempt to manage risk. But it’s not a silver bullet. The model is a tool, and like any tool, its value depends on how you use it.

For now, OpenAI is setting the pace in AI-driven security. Whether that pace is sustainable—or safe—remains to be seen. But one thing is clear: the era of one-size-fits-all AI access is over.

What Happened in Suisun City?

At 5:45 am on August 7, something went very wrong inside Suisun City’s IT network. Malicious software had infected the systems, and within hours, the city declared a state of emergency. The move wasn’t just symbolic — it unlocked access to state and federal resources that smaller municipalities often can’t reach on their own.

The fallout has been wide. The attack disrupted 911 call routing, police and fire dispatch, records management, and general city services. Officials made the call to shut down the entire IT network, a drastic step meant to contain the damage and preserve evidence for a federal investigation.

That decision has real consequences. Online services are offline, City Hall is closed, and in-person meetings across planning, housing, and water departments have been suspended indefinitely.

Is 911 Still Working in Suisun City?

Yes — but not through the usual channels. In an update posted on August 10, the city assured residents that police and fire crews are still responding to emergency calls. Those calls are now being routed through the Solano County dispatch center instead of the city’s own system.

Officials also stressed there is no “imminent” threat to the public from the incident itself. For a city of roughly 30,000 people in Northern California, that’s a small comfort, but a necessary one.

Is This a Ransomware Attack?

Nobody has officially confirmed it yet, but the signs point that way. Suisun City Council Member Princess Washington posted on LinkedIn late on August 10 that an emergency meeting would take place the following day to address the ongoing effects of the incident. The council planned to discuss “threats to public services and facilities, cybersecurity matters and anticipated litigation.”

More tellingly, SFGATE reported that the meeting would include consideration of the city’s response to demands from the “person or persons” behind the malware. When a city starts talking about responding to demands, ransomware is usually the reason.

A Wave of Attacks on Local Governments

Suisun City isn’t an isolated case. It’s part of a troubling pattern that has accelerated over the past few weeks.

• Coweta, Oklahoma (August 5): The city confirmed a “system-wide ransomware attack” and is working with cybersecurity experts to recover systems and assess whether any data was accessed.
• Washburn County, Wisconsin (August 6): Officials issued a press release confirming they’re responding to a cyber incident and shut down technology services as part of the response. No word yet on whether it’s ransomware-related.
• St. Paul, Minnesota (August 2025): The Interlock ransomware group published employee data online after the city refused to pay.
• Clay County, Indiana and Jackson County, Missouri (2024): Both reported ransomware attacks that disrupted critical government services.

These aren’t one-off events. They’re a trend.

Why Are Local Governments Such a Popular Target?

The answer is painfully simple: they’re vulnerable, and attackers know it.

Seemant Sehgal, Founder & CEO of BreachLock, put it bluntly. Municipal IT and security teams, he said, “operate under resource constraints that most enterprise security organizations would find genuinely difficult to imagine.” When three incidents hit in the same news cycle, he added, “it’s clear that attackers have figured that out.”

His assessment is worth sitting with: “Suisun City, Coweta, Washburn County – these are not outliers, they are a pattern.”

Local governments are expected to deliver essential services on tight budgets. Cybersecurity often takes a back seat to roads, schools, and public safety. That calculus is now coming back to bite.

The Cost of a Cyber Incident Goes Beyond Money

For a city like Suisun, the immediate costs are obvious — IT recovery, forensic investigations, potentially a ransom payment. But the hidden costs are just as damaging. Every day City Hall stays closed, permits don’t get processed. Housing meetings get postponed. Water department inquiries go unanswered.

Residents feel the disruption even if their personal data never leaks. And if data does leak, the consequences can linger for years. Just ask St. Paul, where employee information is still floating around on the dark web.

What Can Other Cities Learn From This?

If there’s a silver lining in incidents like these, it’s that they force conversations about preparedness. Cities that haven’t been hit should be paying close attention.

Key takeaways from the recent wave of local government ransomware attacks:

• Have a backup plan for 911 dispatch. Suisun was lucky to have Solano County as a fallback. Not every city does.
• Shut down fast. Taking the entire network offline is painful, but it preserves evidence and limits spread. Hesitation is costly.
• Communicate early and often. Suisun’s updates, while not detailed, at least kept residents informed about what to expect.
• Assume you’re a target. Small cities are not too small to be attacked. They’re often the perfect size — enough data to be valuable, not enough budget to defend it.

The reality is that cyber incidents in government agencies are no longer a matter of if, but when. The cities that recover best will be the ones that planned for that inevitability before the malware hit.

The Discovery: A Decade-Old Hole in Secure Boot

Eleven Microsoft-signed UEFI shim bootloaders are carrying vulnerabilities that let attackers completely bypass Secure Boot. That’s the takeaway from a new ESET report, and the flaws have been sitting there for more than ten years.

These aren’t exotic, hard-to-reach bugs. No memory corruption. No reverse engineering. Just old code that still trusts even older code.

ESET flagged the shims to CERT/CC back in February 2026. All of them are version 0.9 or below, signed under Microsoft’s Microsoft Corporation UEFI CA 2011 third-party certificate. That’s the catch — any system that trusts that certificate will accept these bootloaders, no matter the OS installed.

Why Shims Exist (And Why They’re Dangerous)

A shim is a tiny first-stage bootloader. Microsoft signs it once, and that signature lets Linux distributions boot under Secure Boot without submitting every single update for signing. Convenient, sure. But it also means an attacker can grab a vulnerable shim, carry it to any machine with the Microsoft third-party cert, and boot from it.

The real problem isn’t the shim itself. It’s the second-stage bootloaders these old shims still trust — mostly GRUB 2. The trusted binaries were signed between 2013 and 2025, and older GRUB 2 builds are riddled with well-known flaws.

ESET proved the point using Oracle Linux’s shim. It trusts a GRUB 2 binary vulnerable to a 2015 bug that lets unsigned code load through crafted multiboot modules. The attack is almost trivial: build an unsigned kernel image, drop it next to the old shim and GRUB 2, and load it with a single command at boot.

Bypassing the Defenses Built to Catch This

The shims also sidestep the very mechanisms designed to stop them.

• MOK denylist: Enforcement only arrived in shim version 0.9. Older shims ignore it entirely, so an attacker can load binaries that an organization thought were revoked.
• Secure Boot Advanced Targeting (SBAT): This version-based revocation system came in shim 15.3. Earlier shims never even check the SBAT policy.

That’s a double whammy. Even if you’ve revoked a malicious bootloader, these old shims won’t respect your decision.

The deeper issue is visibility. Shim submissions have only been cataloged transparently since 2017. Nobody knows how many older, still-trusted shims are out there. That’s a scary thought for anyone managing a fleet of Linux servers.

What’s Been Fixed (And What Hasn’t)

Two CVEs cover the reported shims: CVE-2026-8863 and CVE-2026-10797. Microsoft revoked the vulnerable binaries in the dbx update shipped with its June 9 Patch Tuesday.

Windows machines should update automatically. Linux users can pull the revocation through the Linux Vendor Firmware Service.

But here’s the kicker: ESET isn’t releasing indicators of compromise. The vulnerable shims are part of legitimate software packages on thousands of systems that have never been compromised. Releasing IoCs would cause massive misidentification.

Instead, defenders should follow the advice in ESET’s Protection and detection section. That’s the practical path forward.

What This Means for Your Systems

If you’re running Linux on UEFI hardware, this is worth taking seriously. The attack doesn’t need physical access — an attacker just needs to boot from a USB drive or compromise the boot chain.

Check your shim version. If it’s 0.9 or below, you’re exposed. Update through your distribution’s normal channels, and make sure the dbx update is applied.

For security teams, this is a reminder that Secure Boot isn’t a silver bullet. It’s a chain, and every link matters. Old shims are a weak link that’s been hiding in plain sight for a decade.

Related reading: Microsoft fixes 200 CVEs in June Patch Tuesday and how to check your Linux bootloader version.