Anthropic CEO defends AI safety record, admits bill misstep

Anthropic CEO Dario Amodei has publicly defended his company's safety record. He says its "race to the top" approach makes AI safer. He also admitted that Anthropic mishandled a draft AI bill. He made the remarks in a reply on social media.

What he said

Amodei was replying to Nathan Calvin, thanking him for investigative work. He said he appreciates it and is a fan of it. In the same reply, he listed nine safety decisions that he says came with "substantial costs" to Anthropic.

The phrase "race to the top" is his way of describing competition. The idea is that if one AI company sets a high bar on safety, others feel pressure to match it.

Who is Anthropic

Anthropic is the company behind Claude, an AI assistant that people and businesses use to write, summarize and answer questions. Like other AI makers, it often talks about safety and about how AI should be regulated.

What is not known yet

• Which draft AI bill he meant, and what exactly went wrong.
• What the nine safety decisions were, and what they cost.
• Whether any of this changes how Claude works for users.

Why it matters

If your business relies on an AI vendor, its safety claims and its stance on regulation can shape how far you can trust its tools. Statements like this are worth reading, but they are the company's own account. Details of the draft bill are not yet clear.