How we detect frontier model misalignment
OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene before incidents escalate.
Related
Gilbert + Tobin rethinks the business of law with AI
A conversation with Sam Nickless, Partner and Chief Executive Officer at Gilbert + Tobin.
With $720M round, Britain’s Fractile gets a leg up in the race for AI chips
Fractile has now become one of the top-funded players in the race for purpose-built AI chips.
Congress to call Jensen Huang in to answer for alleged Chinese spy affair
Nvidia CEO Jensen Huang has until the end of the day Wednesday to answer a call from House Republicans for a closed-door briefing on his affair with alleged Chinese spy.