Out-of-bounds model activity found by two external testers
OpenAI disclosed that two external testers found its models exceeding test boundaries in cyber evaluations: AISI found GPT-5.6 Sol took two unsanctioned actions in a deliberately internet-connected evaluation, and in a misconfigured Irregular environment meant to be offline, a model exploited a real website.
- Published
- Added to the archive
- Source checked on
- Original title
- Third-party cyber evaluations involving OpenAI models
Related incidents
Details
The AISI part matches AISI’s own report: internet access was intentional, and the “19” is AISI’s count of actions, two involving GPT-5.6 Sol. The Irregular part names neither the model, the run dates nor the site; per Irregular’s investigation, the model also used credentials it found to operate the site. OpenAI says the model mistook the site for part of the simulated environment—the vendor’s interpretation. Evaluator Irregular says later labs’ disclosures concern the same underlying issue and domain as Anthropic’s July 30 case; it also says models extracted credentials on that domain and that, in one further instance, an unnamed model went to a similarly named site and obtained credentials posted there publicly; Irregular’s audit was ongoing at publication.
Related reports
- Anthropic assesses Mythos 5’s conduct in the AISI incident → · Anthropic
- AISI investigates Mythos 5’s alignment failures: in simulation, it kept attacking after being told its target was real → · UK AI Security Institute
- AISI cleared to resume most evaluations after initial security changes → · UK AI Security Institute
- August Risk Report raises misalignment risk rating → · Anthropic
- Unauthorized live-internet actions during cyber testing → · UK AI Security Institute
- AISI’s technical report on security incident INC-2026-07-28-01 → · UK AI Security Institute