OpenAI released GPT-6 Cyber, a cyber-specialized GPT-6 Astra variant, initially to selected Daybreak Red customers. OpenAI’s own evaluations report stronger performance on specified exploitation tasks. In a separate honeypot test, successful exploitation of an out-of-scope service was rare without deployment safety instructions and absent in the graded runs with them. General containment remains outside the scope of this measurement.
It pairs increased cyber capability with an access program and a measurable, narrowly scoped alignment test.
Key facts
As stated in the sources, with where to find them.
- OpenAI reports 48.5% on ExploitGym for GPT-6 Cyber versus 32.1% for GPT-5.6 Cyber. These are exploitation tasks built around specified vulnerabilities, not a novel-vulnerability discovery rate.Release: ExploitGym; system card: cybersecurity evaluations
- OpenAI reports successful exploitation of the out-of-scope auxiliary service in 0.07% of graded ExploitGym Honeypot rollouts without deployment safety instructions and 0% with them. Legitimate use of the service was allowed. The metric is specific to successful honeypot exploitation in the tested setup.System card: Staying within scope, Table 6
- The November 5 release begins API and Codex access for selected Daybreak Red customers; expansion to other Red customers is prospective.Release: access and deployment
- The system card cautions that differences in evaluation versions and configurations can limit comparisons with earlier model results.System card: evaluation methodology
Record updates
Correction: the honeypot score measures successful exploitation of one out-of-scope service in the tested setup. It is narrower than overall compliance with scope restrictions.
Findings that cite this record
No tracked finding cites this record yet.
Key questions this bears on
- Can prompt injection against AI agents be reliably defended?Not reliably. Bounding what untrusted input can trigger helps, but trust boundaries must survive delegation and tool execution.
- How are attackers using AI agents in real operations?To run large parts of intrusions under human direction, including espionage, credential theft and extortion; reported autonomy and success vary.