AI Executives Face Almost No Personal Liability for Safety Failures. One Investor Says That Has to Change
Giorgos Tsetis, the Nutrafol cofounder who has become an active investor in AI companies including Anthropic and xAI, argues that AI executives should face the kind of personal legal liability for safety failures that leaders in industries like supplements already do, rather than treating compliance as a corporate checkbox.
Giorgos Tsetis has spent nearly a decade building Nutrafol, a hair wellness supplement company, into a business where he is personally on the hook if a product causes harm. Now, as an active investor in artificial intelligence companies including Anthropic, xAI, Positron AI, and Lila through his investment effort Great Things, he is arguing that AI executives should face the same kind of personal legal liability for safety failures that supplement industry leaders already live with.
Tsetis's case rests on a simple distinction: corporate penalties, he argues, function as just another line item, a cost that gets absorbed and budgeted for rather than something that changes behavior. Personal liability is different, he says, because it forces executives to proactively identify and address risk themselves rather than treating safety compliance as a checkbox exercise handled by someone else in the organization.
He points to the supplement industry as the model. Leaders there face direct personal legal responsibility when a product causes harm, a standard Tsetis says has no real equivalent in AI today. Executives at AI companies can be held personally accountable for misstatements made to investors, the kind of liability that comes with running any public or well funded private company, but face little comparable exposure for decisions that lead their products or systems to cause harm.
The regulatory landscape is starting to move, even if slowly. Several states have introduced their own AI laws aimed at individual accountability, and Congress is weighing legislation that would impose safety requirements on frontier AI systems. In September, major AI company leaders signed a White House accord committing to internal risk controls, independent audits, and board level safety oversight, and several leading labs have published their own voluntary safety frameworks. All of those measures, however, remain largely corporate commitments rather than the kind of personal legal exposure Tsetis is calling for.
As AI systems take on more consequential decisions in health care, finance, and other high stakes fields, the gap Tsetis is describing, between what executives are liable for today and what he thinks they should be liable for, is likely to keep drawing attention from lawmakers, investors, and the AI companies themselves.
