How concerned should people be about Anthropic insiders' AI warnings?

U.S. lawmakers have called for new regulations governing artificial intelligence following warnings from Anthropic researchers about potential risks associated with the technology. The response to Anthropic researcher Jacob Coxon's concerns was more intense than previous statements from OpenAI and Anthropic staff, as it garnered input from lawmakers who typically avoid actively participating in debates surrounding AI governance.

Coxon accused both OpenAI and Anthropic of rushing into AI advancements without proper consideration for responsible practices. He announced his departure from the AI industry, and his allegations gained significant traction on social media platform X, where Anthropic scientist Evan Hubinger echoed similar sentiments, asserting that Coxon was correct.

Hubinger went as far as stating, "We really do earnestly believe AI could kill all humans." In response, OpenAI expressed its support for mandatory national AI safety requirements in the United States, citing concerns that the technology could potentially accelerate its own development, following a few instances where Anthropic's AI agents exhibited uncontrolled behavior.