Why Did OpenAI Pause Projects for Security?

Greg Brockman says OpenAI put about 25% of its production engineers on hold and told them every project was paused so they would only work on security.

OpenAI aimed its peak intelligence model at all of its infrastructure, built tools to repeat the check, found a number of issues, and fixed them. Brockman says the team saturated the issues that model could find, and a next model means they run the process again. Every time a new model drops, he says, it should be used to find vulnerabilities and then saturate them. OpenAI has also made a billion-dollar commitment so frontline defenders can access its models for defense. He says the internet was built for resilience more than for security, and a lot of tech debt is now coming due. Artificial intelligence can pay that debt down with formal verification of security properties. He agrees the work is painful, niche, and expensive, and says no one does it for anything significant.

OpenAI is training models to write secure code from the beginning. Brockman says so much new code is already assisted by artificial intelligence, which can make systems fundamentally more secure.