The US artificial intelligence (AI) firm Anthropic is looking to hire a chemical weapons and high-yield explosives expert to try to prevent “catastrophic misuse” of its software.
In other words, it fears that its AI tools might tell someone how to make chemical or radioactive weapons, and wants an expert to ensure its guardrails are sufficiently robust.
In the LinkedIn recruitment post, the firm says applicants should have a minimum of five years experience in “chemical weapons and/or explosives defence” as well as knowledge of “radiological dispersal devices” – also known as dirty bombs.
The firm told the BBC the role was similar to jobs in other sensitive areas that it has already created.
Anthropic is not the only AI firm adopting this strategy.
A similar position has been advertised by ChatGPT developer OpenAI. On its careers website, it lists a job vacancy for a researcher in “biological and chemical risks”, with a salary of up to $455,000 (£335,000), almost double that offered by Anthropic.
But some experts are alarmed by the risks of this approach, warning that it gives AI tools information about weapons – even if they have been instructed not to use it.
“Is it ever safe to use AI systems to handle sensitive chemicals and explosives information, including dirty bombs and other radiological weapons?” said Dr Stephanie Hare, tech researcher and co-presenter of the BBC’s AI Decoded TV programme.
“There is no international treaty or other regulation for this type of work and the use of AI with these types of weapons. All of this is happening out of sight.”
The AI industry has continuously warned about the potential existential threats posed by its technology, but there has been no attempt to slow down its progress.
The issue has gained urgency as the US government calls on AI firms while launching war in Iran and military operations in Venezuela.
Anthropic is taking legal action against the US Department of Defence, which designated it a supply chain risk when the firm insisted its systems must not be used in either fully autonomous weapons or mass surveillance of Americans.
Anthropic co-founder Dario Amodei wrote in February that he didn’t think the technology was good enough yet, and should not be used for these purposes.
