The Anthropic CEO says they will «work to enable a smooth transition,» after denying the US military use of their AI for mass surveillance or autonomous killing.
— In a narrow set of cases, we believe AI can undermine, rather than defend, democratic values, writes Amodei, — Some uses are also simply outside the bounds of what today’s technology can safely and reliably do.
Committed to U.S. leadership
He says that Anthropic is committed to American leadership in AI, having refused «several hundred million dollars» in Chinese government-linked contracts, denying CCP-sponsored cyberattacks and argued for export controls on chips.
The two things Anthropic deny doing are because AI can quickly and easily put together profiles of the entire life of people through public sources, and he says laws lag behind capabilities.
Autonomous killings are just plain too dangerous, he argues, saying that today’s AI systems «are simply not reliable enough to power fully autonomous weapons.»
«Supply chain risk»
The Pentagon’s next move could be to blacklist the company as a «supply chain risk,» which would also entail that all government contractors would have to consider their contracts, Axios writes.
They could also use the Defense Production Act to compel compliance from Anthropic.
These two moves are contradictory, Amodei says. One is banning or replacing them for not being compliant enough, while the other says that Anthropic is so valuable they cannot be ignored.
— Regardless, these threats do not change our position: we cannot in good conscience accede to their request, Amodei closes.
Read more: Anthropic’s response, Axios, The Verge, Reuters.
One thought on “Amodei officially says Anthropic won’t drop Pentagon safeguards”
Comments are closed.