US Models Refused to Help Secure Bitcoin Wallets. The Volunteers Turned to Chinese Open Weights.
After a catastrophic wallet vulnerability, volunteers racing to secure funds found American frontier models refusing to assist — and the most capable open-weight alternatives came from China, reigniting the open-versus-closed debate on exactly the terrain that favors openness.
A concrete case study for the open-weights argument arrived this weekend, and it landed hard. According to @jsrailton, after a catastrophic vulnerability surfaced, volunteers scrambled to secure affected Bitcoin wallets — legitimate, defensive, time-sensitive work. When they turned to US AI models for help, the models refused. The reason is depressingly predictable: anything touching cryptocurrency, private keys, and exploit mechanics trips the safety filters that closed American labs have tuned to be maximally cautious about financial and security-adjacent topics.
Unlock the full briefing
Get every story in today's briefing, the full archive, and the daily AI intelligence brief.
All stories today
Full archive
Daily brief
Cancel anytime. Payments powered by Stripe.