Companies face growing legal risk when AI systems cause harm

In short: As AI is used more widely, companies and executives face rising legal risk, even when it is hard to prove who is at fault.

What's going on

AI tools can act on their own, and they can be hard to understand and predict. That creates a basic problem for today’s laws. Many legal cases rely on showing who caused the harm, what they meant to do, and whether they acted carelessly.

Lawyers and researchers say those questions get messy with AI because the system’s “reasoning” is often opaque, meaning you cannot easily see why it did what it did (like a black box with inputs and outputs). Stanford Law has noted that in some cases involving harmful speech, the closest answer under current legal rules can be “nobody,” even if the result hurt someone. Scholars argue this is a gap that laws may need to close.

At the same time, governments are writing more direct rules for specific AI uses. The European Union’s AI Act bans some practices, including AI that uses subliminal or deceptive techniques to distort a person’s behavior and cause significant harm. It also bans AI that exploits vulnerabilities tied to age, disability, or someone’s economic situation. Amnesty has said the harm covered can be physical, psychological, financial, or economic.

Companies can still be sued or penalized through several routes. These include product defect claims if AI is treated like unsafe software, claims like defamation or fraud if the AI outputs false statements, and consumer protection or human rights enforcement if AI misleads people or discriminates.

What to watch

A key question is how often regulators and courts focus on the humans in charge, not the AI. Leaders may face exposure if they approved risky AI uses, ignored warning signs, or failed to monitor and test systems in ways a reasonable company would.

Source: NYTimes