OpenAI briefs government agencies and Five Eyes on GPT-5.4-Cyber as AI cybersecurity race heats up

OpenAI Engages Government Agencies with New Cyber Model

OpenAI held a demonstration event in Washington for approximately 50 cyber defense practitioners across federal government agencies to showcase its new GPT-5.4-Cyber model, which rolled out under a tiered access program 2

2

. The briefing included officials from across the government and national security agencies, most of whom oversee day-to-day cyber tasks . OpenAI Chief Global Affairs Officer Chris Lehane explained that this approach would allow more organizations, including local water utilities, to access advanced AI tools for combating cyber threats 2

2

.

The company is pursuing a dual-track approach, making one version of its model more widely available with strong safeguards in place, while releasing another more cyber-permissive version to defenders through the Trusted Access program 2

2

. Government applicants are going through the same vetting process as commercial customers who wish to join this program 2

2

. Sasha Baker, OpenAI's head of national security policy, told attendees that OpenAI hopes to partner with government departments to prioritize crucial use cases and build channels to share threat intelligence across sectors 2

2

.

Five Eyes Intelligence Alliance Gains Access

OpenAI is starting briefings with Five Eyes intelligence alliance members this week to get them vetted and signed up to access the model 2

2

. In addition to the U.S., that intelligence-sharing partnership includes Australia, Canada, New Zealand and the U.K. 2

2

. The company is also working with state governments to get them access to GPT-5.4-Cyber 2

2

. Most companies that already have access to OpenAI's model are using the tools to find exploitable security vulnerabilities in their own internal systems 2

2

.

OpenAI unveiled GPT-5.4-Cyber earlier this month as a variant of its latest flagship model fine-tuned specifically for defensive cybersecurity work 3

3

. The company has supported defenders since 2023 through its Cybersecurity Grant Programme, spending $10 million in API credits to accelerate adoption among security teams, particularly those focused on identifying and fixing software flaws 3

3

.

Competition with Anthropic Mythos Preview Intensifies

OpenAI rolled out its new cyber model hot on the heels of Anthropic Mythos Preview, and both companies are currently working with government agencies to determine who will have access 2

2

. Anthropic withheld a public release of Mythos, citing its cyber risks, and offered it only to around 40 companies and organizations, including at least two in the federal government 2

2

. Meanwhile, OpenAI is capitalizing on the confusion and quickly engaging with federal, state and international government offices to deploy its competing product 1

1

.

Anthropic's rollout within the U.S. government is complicated by the Pentagon's decision to label the company as a supply chain risk after a messy AI safeguards fight 2

2

. Still, Mythos is currently being tested by the National Security Agency (NSA) despite the designation 2

2

. The White House is still negotiating access to Anthropic's Mythos Preview, which the company has held back due to its hacking capabilities 1

1

. Only a patchwork of agencies—including the NSA and the Department of Commerce's AI testing institute, but not CISA—have access to the model 1

1

.

U.S. Cyber Command Builds Model-Agnostic Infrastructure

U.S. Cyber Command is largely sidestepping the debate between AI models by building infrastructure designed to swap between AI models regardless of vendor or origin 1

1

. "To survive anywhere, just in case our operators want an open-source made-in-China model or something very boutique, we have to create the infrastructure and that ability to be agile—no politics," Brig. Gen. Reid Novotny, chief AI officer at Cyber Command, told Axios 1

1

.

2026 marks the first year Cyber Command has dedicated funding for AI programs, after years of lead time inside the Pentagon and Congress 1

1

. The command is using that funding to pilot commercial AI capabilities while building underlying infrastructure that allows operators to switch between models as technology evolves 1

1

. Novotny said that flexibility extends even to models developed outside the U.S. 1

1

.

A major concern externally is whether AI systems could misidentify or improperly target critical infrastructure. But Novotny said those risks are governed by existing military rules, not new AI-specific policies 1

1

. Cyber Command is testing varying levels of human oversight, but fully autonomous deployment is off the table 1

1

. "We would never unleash a human-out-of-the-loop tool and then be like, 'Oops, we just turned something on,'" Novotny said 1

1

.

Microsoft Partnership Strengthens AI Cybersecurity Efforts

OpenAI and Microsoft announced an expanded cybersecurity partnership to deploy advanced artificial intelligence tools against cyber threats 3

3

. OpenAI will provide Microsoft access to its most advanced cyber-capable AI models through its Trusted Access for Cyber programme 3

3

. Microsoft, in turn, will apply its cybersecurity infrastructure and expertise, including its Secure Future Initiative, to help protect OpenAI's systems, models and shared customers 3

3

.

Microsoft recently evaluated Mythos using its own open-source benchmark for real-world detection engineering tasks, and the results showed substantial improvements relative to prior models 3

3

. OpenAI has expanded tooling such as Codex Security, which automatically scans codebases, validates vulnerabilities and proposes fixes 3

3

. According to the company, the system has contributed to fixing more than 3,000 critical and high-severity vulnerabilities since its recent rollout 3

3

.

Implications for Cyber Warfare Capabilities

AI's role in cyber warfare is expanding as vendors race to secure Pentagon contracts and adversaries like China integrate AI into their operations 1

1

. Google has signed a deal with the Pentagon to use Gemini in classified government operations, according to The Information 1

1

. Anthropic was first into the classified space, followed by OpenAI earlier this year 1

1

.

Government officials have raised concerns that artificial intelligence tools could be misused to disrupt critical infrastructure such as financial systems or power grids 4

4

. Former military leaders and industry operators say the bigger challenge is how quickly the command can put AI models to use 1

1

. Since many government agencies contend with outdated computing systems that are difficult to secure, these AI tools are expected to help significantly accelerate the process of identifying critical weaknesses before it is too late 5

5

.