Anthropic Hires Chemical Weapons Expert to Prevent Claude Misuse; Experts Question Increased Risk

Anthropic recently posted a 'Chemical Weapons and High-Yield Explosives Policy Manager' position on LinkedIn, requiring applicants to have at least 5 years of relevant defense experience. The role aims to prevent its AI tool Claude from being misused for catastrophic purposes, such as obtaining information to make chemical or radiological weapons. This position is similar to other sensitive roles the company has set up.

OpenAI is also hiring a biological and chemical risk researcher with a maximum annual salary of $455,000, nearly twice that of Anthropic's position. Dr. Stephanie Hare, a tech researcher, questions whether handling such sensitive information is safe even if AI is instructed not to use it; there are currently no international treaties regulating the combination of AI and such weapons, and related work is done outside public view.