OpenAI Launches New Team to Assess and Mitigate Risks of Advanced AI

OpenAI created a new team called Preparedness to assess and protect against risks from future AI systems. The team will look at threats like AI's ability to fool humans and generate malicious code.

Preparedness will also study more speculative threats like chemical, biological, radiological and nuclear risks from AI models.

OpenAI is soliciting ideas for risk studies from the community, with prizes for the top submissions.

The Preparedness team will formulate policies and tools for risk monitoring, mitigation, and governance during AI model development.

This effort complements OpenAI's other work in AI safety, focused on risks from highly capable AI systems that could exceed human intelligence.