Crypto Pioneers Sound Alarm Over Government Control of AI Knowledge

Crypto figures are sounding the alarm about government control over AI knowledge, warning that it could create a dangerous precedent. According to Erik Voorhees, founder of ShapeShift, states should not decide what forms of intelligence are 'safe.' This concern was sparked by Anthropic's statement on open-weights models, which CEO Dario Amodei said his company has never advocated banning. Instead, the main concerns were about authoritarian governments developing advanced AI for military or surveillance uses, as well as misuse in cyberattacks or biological threats.

Voorhees argued that allowing governments to define acceptable AI knowledge could expand beyond its original purpose, potentially leading to a slippery slope where more and more information is restricted. He used a hypothetical progression to illustrate this point, starting from a seemingly reasonable rule against discussing bioweapons and ending with the banning of unapproved encryption and a requirement that AI not obstruct government orders.

This debate has been ongoing, with some experts proposing federally backed bodies to test and certify frontier models before release. However, others argue that existing regulators and laws are sufficient, and adding more approval processes could be counterproductive. The issue highlights the tension between ensuring AI safety and preserving intellectual freedom.