BREAKING
OpenAI Opens GPT-5.5-Cyber Fully
Trusted Access for Cyber
Three Access Tiers
GPT-5.5 default
Tier 1
●
Standard safeguards
GPT-5.5 with TAC
Tier 2
●
Secure code review
●
Vulnerability triage
GPT-5.5-Cyber
Tier 3
●
Authorized red teaming
●
Exploit validation
0
%
AISI expert pass rate
Fixing, Not Just Finding
1
Threat model
↓
2
Explore attack paths
↓
3
Validate in sandbox
↓
4
Propose patch
Power vs. Access Controls
AI NEWS BLITZ
OpenAI has rolled out the full GPT-5.5-Cyber model to vetted defenders.