Elon Musk publicly admitted he was wrong about Anthropic. The company is now the leader in AI, he says — no one has released better models than Mythos and Fable.
https://twitter.com/elonmusk/status/2075278580955685036?s=46
That's a pretty sharp U-turn.
Back in February, Musk was calling Anthropic "anti-Western civilization," claiming its models hate white people, Asian people, heterosexuals, and men. He called the company "anti-human and evil." He accused it of stealing training data on a massive scale and described it as "self-righteous, sanctimonious, and hypocritical." Come March, he was still asking: "Is there a more hypocritical company than Anthropic?"
The turning point came after they started doing business together.
In May, Anthropic leased SpaceX's Colossus 1 data center, which comes with access to over 220,000 Nvidia GPUs. Musk then said he'd spent a lot of time talking with Anthropic's leadership and found them "very capable and genuinely concerned about doing the right thing." He added that none of them triggered his "evil detector."
Two months later, that evil detector has been upgraded into a championship certification. Not only does Musk now rank Anthropic as the #1 AI company, he also promised he won't suddenly cut off their compute power to hurt them — because "that's not my style."