OpenAI introduced GPT-5.6-Cyber on Monday, August 10, a specialized artificial intelligence model designed for advanced, authorized cybersecurity work. The company is making it available only through the new Daybreak Red access tier for approved defenders, as concern grows that attackers will increasingly automate cyber operations. Axios independently reported the launch at 17:00 UTC, placing the announcement within the preceding 24 hours.
The model is built on GPT-5.6 Sol but is trained to handle higher-risk, dual-use tasks with fewer refusals. OpenAI said Daybreak Blue will give approved users access to the general model with safeguards tailored for defensive work such as incident response, malware analysis and patch validation. Daybreak Red is intended for authorized vulnerability research, exploit validation, penetration testing and other advanced security assessments.
In an internal evaluation covering exploit-chain development, authentication bypass and privilege escalation, OpenAI said GPT-5.6-Cyber completed 95.0% of requests. That compared with 1.5% for the standard GPT-5.6 Sol and 2.0% for Sol under Daybreak Blue. The earlier GPT-5.5-Cyber completed 57.3%. Those figures are company results and have not been independently reproduced.
OpenAI also said the model helped identify two previously unknown flaws in Google's V8 JavaScript engine that could be chained to corrupt memory and escape its heap sandbox. Google fixed the reported issue and assigned CVE-2026-15903. The company said the model found additional high-severity issues in a mobile operating system, a database and an operating-system kernel, but details remain withheld while remediation and coordinated disclosure continue.
The wider access comes with controls. OpenAI said both tiers require approval, identity verification, account security, monitoring, limits on permitted uses and legal attestations. Individual Daybreak accounts must adopt hardware security keys from September 1. The company also recommends isolated environments, explicit authorization boundaries, human oversight and an auto-review mode that can block potentially destructive actions requiring elevated permissions.
OpenAI assessed GPT-5.6-Cyber at the High cybersecurity capability level under its Preparedness Framework, below the Critical threshold. It acknowledged that reduced safeguards create risks from misuse or misalignment, and said a fuller system card will follow. The launch highlights the industry's difficult balance: giving legitimate defenders more capable tools before offensive AI spreads, while limiting access tightly enough to prevent the same capabilities from empowering attackers.
Comments