Greg Brockman: OpenAI training models for superhuman secure code

AccBalanced · x · 2026-08-18

OpenAI President Greg Brockman stated that the company is developing models capable of writing "superhumanly secure code" to harden infrastructure against AI-powered attacks. He argued that organizations must significantly automate security programs soon, keeping human review for consequential changes but using AI to eliminate delays between identifying issues and deploying fixes.

Related event: OpenAI Trains Models to Write Superhumanly Secure Code(2 posts)→

Original post →

More from Companies & People

Companies & People channel →