US AI liability gap leaves 1,200 lawsuits in legal limbo

The Altitudes piece captures a core asymmetry of the AGI race: technical capability is scaling quickly, while US liability law remains stuck in a pre‑AI era. With no federal statute defining who is responsible when an AI system harms someone, courts are improvising with product liability, negligence and First Amendment doctrine—and reaching conflicting conclusions on nearly identical fact patterns. That legal uncertainty effectively outsources AI governance to whichever judge someone happens to draw.

For frontier labs, the regulatory vacuum is a double-edged sword. In the short term it means fewer hard constraints on deploying increasingly capable models into sensitive domains like healthcare, finance and employment. In the longer term it raises tail risks: one catastrophic failure could trigger a political backlash that forces much more draconian rules than a measured liability framework would have. It also advantages the largest players; they can afford bespoke legal teams and insurance, while smaller AI companies struggle with fragmented state requirements.

From an AGI-timeline perspective, the lack of clear liability probably nudges deployment faster than it otherwise would, especially in high-risk, high-reward sectors. But it also increases the chance that when something goes wrong at scale, the policy response will be blunt and reactionary rather than calibrated.