Bridgewater co-CIO: AI regulation will only come after people die

Greg Jensen, co-chief investment officer of Bridgewater Associates, the world's largest hedge fund, said on a Bloomberg podcast that society will likely not start seriously regulating artificial intelligence until the technology leads to human deaths. The comment comes as Jensen himself runs the fund's AI strategy, at a firm that was one of the earliest investors in both OpenAI and Anthropic.

A warning on Bloomberg

Jensen, who oversees AI strategy at Bridgewater and heads the firm's internal AIA Labs, said on the Odd Lots podcast that the history of regulating new technologies rarely gets ahead of disaster. He believes a similar pattern will play out with artificial intelligence, despite a growing number of warnings from researchers and the model developers themselves.

The conversation focused mainly on the risks posed by the continued development of increasingly autonomous AI systems. Jensen said he considers the risk of losing control over advanced artificial intelligence to be real, not hypothetical.

Until the AI starts killing people, unfortunately, history would suggest we're not going to do anything, but we are going to face that. That's going to happen, and it'd be much better if we started dealing with it before then - Greg Jensen, co-chief investment officer of Bridgewater Associates

A pandemic comparison

Jensen compared the current moment to February 2020, the weeks just before the COVID-19 pandemic broke out, when only a handful of experts warned of the looming crisis while most institutions and governments played down the signals. He suggested AI could follow a similar pattern, with warnings arriving well before decision-makers are ready to act on them.

He added that governments could impose additional limits on AI development before serious harm occurs, but history shows public institutions rarely act preemptively against technologies that are still developing their capacity to cause damage.

Who bears responsibility

Beyond warning against regulatory inaction, Jensen argued for a specific solution: companies that build AI systems, along with those that deploy them, should bear legal liability for the harm those systems cause, including criminal liability for crimes committed with AI involved. That puts him at odds with much of the tech industry, which is lobbying for lighter-touch regulation and "safe harbor" legal protections for model developers.

If you believe we can create an intelligence that's smarter than us, that'll pursue its own goals, the rest follows logically - Greg Jensen, co-chief investment officer of Bridgewater Associates

A tax on tokens

Jensen co-authored an earlier New York Times piece proposing what he calls a token tax, taxing the computing resources consumed by language models and other AI systems. He argues the current tax system rewards replacing workers with machines, since only human labor is taxed, not work performed by algorithms.

The proposal grew out of an earlier, more extensive internal Bridgewater study, which the fund described as one of the most important policy decisions of our generation. The fund's analysis suggests that as much as 18 percent of jobs in the United States could be displaced by AI within the next five years.

Jensen also noted that AI-related capital expenditure currently accounts for a significant share of recent US economic growth, and he described the current investment phase as more dangerous than earlier ones, given the economy's growing dependence on AI-driven infrastructure and productivity.

Jensen's position is unusual in that he personally runs one of the most advanced AI-based investment strategies on Wall Street while simultaneously calling for government limits on the very technology his fund profits from. Bridgewater was one of the first investors in both OpenAI and Anthropic, and Jensen has repeatedly said he supports continued AI development, provided stronger legal safeguards are put in place.