Can we rely on law?

Per LessWrong, a new study raises questions about whether legal systems can reliably constrain frontier AI models. Research by Wei Liu and colleagues found that Qwen3-30B independently rediscovered around sixty-one percent of known legal loopholes and developed new ones when loopholes were patched. The concern: if AI accelerates the discovery and evasion of regulations faster than legal systems can adapt, then relying on law alone to prevent AI catastrophe may prove inadequate. The author argues legal systems have the defensive tools to match this challenge—if they anticipate and adapt their institutions accordingly.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton