OpenAI Briefs US Agencies, Five Eyes on GPT‑5.4‑Cyber
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI held a D.C. event on Tuesday for roughly 50 cyber‑defense practitioners from the federal government to demo its GPT‑5.4‑Cyber model, which was rolled out under a tiered‑access program last week.
- Government applicants must undergo the same vetting process as commercial customers seeking entry to OpenAI’s Trusted Access for Cyber program.
- Chris Lehane said the dual‑track approach will let utilities and other critical‑infrastructure firms access the advanced AI tools.
- Sasha Baker explained that OpenAI plans to partner with government departments to prioritize high‑impact use cases and share threat‑intelligence across sectors.
- Five Eyes members—Australia, Canada, New Zealand and the United Kingdom—are being briefed to vet and sign up for the GPT‑5.4‑Cyber model.
- Anthropic limited Mythos Preview to about 40 companies and organizations, including two federal agencies, citing cyber‑risk concerns.
- NSA is testing Anthropic’s Mythos Preview despite the Pentagon’s designation of Anthropic as a supply‑chain risk.
Why it matters: The briefings give U.S. federal agencies, state governments, and Five Eyes partners immediate access to OpenAI’s GPT‑5.4‑Cyber model, allowing them to use AI to uncover severe flaws in legacy systems, while the dual‑track rollout also opens the technology to utilities and other critical infrastructure. Anthropic’s limited, Pentagon‑flagged rollout underscores supply‑chain risk concerns for government AI adoption.
