Press Space to continue
Press Space to continue
Ex-OpenAI safety researcher @sjgadler reveals the Guidelight control scorecard grades: OpenAI and Anthropic tied for first at C+, Google at D+, SpaceXAI at D-, and Meta at F on basic safety practices "OpenAI and Anthropic tied for first place in our assessment. They got a C+. Google came middle of the pack at a D+. SpaceXAI and Meta are even below that. They're respectively a D- and an F." "We go through different practices, like how well can the company see what their AI is doing when they…
sjgadler@sjgadlerAug 18
Our team spent hundreds of hours reading documents so you don’t have to, all to answer: How good are AI companies’ safety practices? I’m really proud of what we’ve built: It’s Guidelight’s first scorecard, on whether companies can control their AIs, and it's launching today. https://t.co/EQ1Ql1h6LQ
View