Robert Reich urges using liability law to curb AI and climate risks

Reich’s column is not new law, but it captures an emerging mood: if legislatures move slowly on AI, plaintiffs’ lawyers and courts may not. By explicitly linking AI to climate in the frame of “existential risk”, he is teeing up a narrative where major AI developers could face class actions and public nuisance suits similar to those brought against fossil fuel companies. Even if the legal theories are untested, the threat of discovery, injunctions and large settlements can be enough to shape corporate behavior.

For AGI, the prospect of liability-driven governance is a double-edged sword. On the one hand, the possibility of being held financially responsible for large-scale harms could push labs to take safety, red teaming and deployment controls more seriously than any voluntary commitment. On the other hand, poorly targeted lawsuits could chill beneficial research, concentrate the field in a few firms that can afford the legal risk, or drive activity into jurisdictions with weaker legal systems.

This piece is also part of a broader trend: traditional legal tools are being retrofitted to AI instead of waiting for bespoke AI statutes. That suggests the practical constraint on frontier experimentation may come from judges and insurers long before a comprehensive AI act clears Congress.