Top hacker at Anthropic is blown away by their own model: from blocking releases to becoming a White House lobbyist
Nicholas Carlini was once the most notorious skeptic in the AI security scene. While at Google, he publicly mocked OpenAI for delaying the release of GPT-2, calling them 'too conservative' due to safety concerns. Now, this 35-year-old hacker is Anthropic's top security expert—after being impressed by his own model's capabilities, he has shifted from blocking releases to advocating for AI safety policy at the White House.
Why it matters: This marks a shift in the mindset of top AI security experts—when model capabilities exceed expectations, there needs to be a new balance between technical safety and policy advocacy.
#Anthropic #AI #安全 #WhiteHouse
Nicholas Carlini was once the most notorious skeptic in the AI security scene. While at Google, he publicly mocked OpenAI for delaying the release of GPT-2, calling them 'too conservative' due to safety concerns. Now, this 35-year-old hacker is Anthropic's top security expert—after being impressed by his own model's capabilities, he has shifted from blocking releases to advocating for AI safety policy at the White House.
Why it matters: This marks a shift in the mindset of top AI security experts—when model capabilities exceed expectations, there needs to be a new balance between technical safety and policy advocacy.
#Anthropic #AI #安全 #WhiteHouse