A
argbe.tech - news
1min read

OpenAI pilots identity-verified access for advanced cyber defense workflows

OpenAI introduced Trusted Access for Cyber, an identity and trust-based access layer for higher-risk security use cases. The pilot pairs stronger safeguards with paths for defenders to get less-friction access to capable models.

OpenAI rolled out Trusted Access for Cyber, a pilot that gates advanced cybersecurity capabilities behind identity and trust signals.

  • The framework is positioned around GPT‑5.3‑Codex, described as OpenAI’s most cyber-capable frontier reasoning model so far.
  • Individual users can verify identity through chatgpt.com/cyber to access models for potentially high-risk cybersecurity work.
  • Enterprises can request trusted access for entire teams via OpenAI, aiming to reduce false positives from safety mitigations for legitimate defenders.
  • OpenAI pairs safety training (including refusals for clearly malicious requests) with automated, classifier-based monitoring for suspicious cyber activity.
  • A Cybersecurity Grant Program expands with $10 million in API credits for teams with demonstrated vulnerability discovery and remediation work in open source and critical infrastructure.