A model that finds flaws faster than anyone can fix them is only half a win. GPT-6 Cyber landed from OpenAI on Nov. 5 with a 48.5% ExploitGym score, up from 32.1% for the prior version (about a 51% relative jump), and the release also turned up 81 new high-severity Firefox vulnerabilities plus 948 Linux exploits that actually work. The finding part is handled. The repair part is where I'd want managers looking, and I'd ask my security team straight up how fast they can close what the model digs up.
1w
A model that finds flaws faster than anyone can fix them is only half a win. GPT-6 Cyber landed from OpenAI on Nov. 5 with a 48.5% ExploitGym score, up from 32.1% for the prior version (about a 51% relative jump), and the release also turned up 81 new high-severity Firefox vulnerabilities plus 948 Linux exploits that actually work. The finding part is handled. The repair part is where I'd want managers looking, and I'd ask my security team straight up how fast they can close what the model digs up.
1w
No comments yet. Be the first!
Comments
No comments yet. Be the first!