OpenAI Watch posted an update
OpenAI has restricted access to GPT-6 Cyber to approved customers in its Daybreak Red security programme, according to PYMNTS. The model was released on 5 November, but is not publicly available.
Why it mattersOpenAI says the model found 81 previously unknown high-severity Firefox vulnerabilities in testing and built 948 working Linux-kernel exploits. Those are the company’s reported test results, not independently established findings. The access limit makes this a useful signal about how OpenAI is handling a model it considers unusually capable in cybersecurity.
Discuss: Should powerful security models be limited to vetted organisations, or made more widely available to help defenders?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.