最新 Anthropic 新聞與觀點 | Yellow.com

Anthropic has reportedly finished training a more capable successor to Mythos 5, just 9 days after U.S. export controls forced the model offline worldwide. Key Points: A new Mythos model has reportedly cleared training, though its name and release plans stay unknown. The work continued despite a U.S

President Donald Trump said Anthropic no longer looks like a national security threat after meeting CEO Dario Amodei at the G7 summit. Key Points: Trump told Axios that Anthropic had “behaved very responsibly” after weeks of pressure from his administration. The Commerce Department’s Jun. 12 directi

Researchers at OpenAI say reinforcement learning aimed at beneficial traits can broadly improve AI behavior, with gains that spread to new domains and hold under adversarial pressure. OpenAI Trait Training The findings appear in a paper published Jun. 18. Its correspondence authors, Akshay V. Jagade

President Donald Trump on Friday stated that Anthropic does not pose a national security threat and described the company as behaving responsibly. The comment came weeks after reporting suggested the administration had forced Anthropic to withdraw a model access proposal under government pressure. A

Dario Amodei used a Jun. 17 G7 lunch with Donald Trump to call for AI unity, days after the U.S. restricted Anthropic models. Key Points: Amodei urged democratic governments not to split over access to frontier AI systems. The U.S. Commerce Department restricted Anthropic’s Fable 5 and Mythos 5 mode