OpenAI limits GPT-5.5-Cyber to verified defenders — Japan among seven government partners
OpenAI moved GPT-5.5-Cyber from preview to full release under its Daybreak initiative, posting 85.6% on CyberGym and automating the full vulnerability-to-patch pipeline across large codebases — and is distributing the model exclusively to verified trusted defenders, not the open API. Japan is one of seven government-level Trusted Access partners — alongside South Korea, Australia, Canada, Germany, France, and EU institutions including ENISA — eligible to apply via a government-to-government channel, positioning OpenAI's most capable cyber model inside state trust structures rather than commercial ones.
Source: cybersecuritynews.com ↗
The historical bottleneck of finding vulnerabilities has given way to a new challenge: patching them at scale.
Why this matters
- → AI-automated vulnerability patching at scale shifts cybersecurity bottleneck from detection to remediation.
- → State-restricted access concentrates advanced cyber AI inside government trust structures, not commercial mark
- → 85.6% benchmark performance enables end-to-end vulnerability-to-patch pipeline across massive codebases.