Anthropic CEO Amodei warns on open-source AI after Kimi K3
Anthropic's CEO, Dario Amodei, is weighing in on the big debate about open-source AI.
After a Chinese model called Kimi K3 beat several US models, sparking talk of sanctions from the Trump team, Amodei says he doesn't want to ban open-source models but thinks we need to be careful.
Meanwhile, tech giants like NVIDIA and Microsoft are pushing back against sanctions, saying open-source AI helps drive innovation.
Amodei warns open-source AI risks misuse
Amodei is mainly concerned that open-source models could boost China's military power or be misused for cyber and biological attacks.
He's suggesting some guardrails: mandatory safety testing for all AI models open or closed, tighter controls on large-scale copying of AI tech, and stopping US chip exports to China.
NVIDIA says open-source helped stop attack
NVIDIA argues that open-source AI can actually help with cybersecurity: they pointed out a recent case where Chinese open-source technology stopped a rogue agent attack.
Still, tensions are high: an NVIDIA employee was just detained in Taiwan for allegedly smuggling AI chips to China.