Anthropic Refuses to Sign the Open Letter on Open Weights Models

A writer at AI StopWatch describes how an open letter opposing bans on open weights AI models pressured major companies into signing, praises Anthropic's chief executive for declining and for his reasoning about misuse risks, and still argues that the broader race toward superintelligence must be halted.

An open letter opposing bans on open weights AI models put public pressure on major companies, and Google and OpenAI both signed it.
Anthropic's chief executive declined to sign and published a blog post saying his company has never called for banning open weights models, which he considers a public good when they lack dangerous capabilities.
His concern is that powerful models could aid cyberattacks or biological attacks, and that once weights are released they cannot be recalled, monitored, or reliably kept behind safety guardrails.
He proposes halting sales of powerful AI chips to China, restricting the copying of frontier capabilities into less protected models, and requiring safety testing for every model, whether open or closed.
The author supports that refusal but disagrees that frontier development is safe with testing alone, and says the race to superintelligence should be stopped.