OpenAI unveils cybersecurity model amid concerns over AI-driven exploits

Max Justicz, CTO and cofounder of Charge Robotics, used OpenAI’s GPT 5.6 Sol through the Daybreak Blue cybersecurity program to identify a severe vulnerability in code used by Ubuntu. The flaw could have enabled attackers to take control of affected servers. Justicz reported it to the Debian Project, which maintains the software and released a patch. No evidence of exploitation is known.

The discovery took about two hours, according to Justicz, who had previously found a vulnerability in the same codebase. He said the result showed how powerful models could help uncover difficult bugs, while also raising concerns that less-experienced attackers might use similar tools. Justicz estimated that he spent about $5,000 on subscriptions and tokens while finding several exploits.

OpenAI has introduced ChatGPT Cyber for the highest-tier Daybreak Red users, who conduct offensive security research. The company said the model helped identify 948 potential Linux kernel issues and 81 novel Firefox vulnerabilities. OpenAI also announced a $1 billion cybersecurity initiative. The company says only thousands of professionals participate in Daybreak, while researchers question whether defenders have sufficient access and resources compared with potential attackers. OpenAI said it is directing substantial frontier computing capacity toward finding and patching open-source vulnerabilities for free.