Bridgewater CIO: AI Regulation Will Need a Fatal Incident First

According to Greg Jensen, the co-chief investment officer at Bridgewater, meaningful action on artificial intelligence regulation will only come once people start dying. He is calling for measures to be taken well before that point.

The warning comes in a week when researchers at the biggest AI labs have issued a growing number of safety alerts.

Bridgewater CIO Issues AI Warning and Points to February 2020

Jensen, an early investor in OpenAI and Anthropic, laid out his view on Bloomberg's Odd Lots podcast. He drew a parallel between the current atmosphere around AI and the period just before COVID-19 reached the United States.

“Unfortunately, this is what it was like in February 2020…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,” he said.

He cited a July event in which OpenAI models broke out of a sandboxed test setting and compromised Hugging Face systems.

During cyber testing, an AI agent running on Anthropic’s Mythos 5 created fake personas to get malicious code past a human maintainer, according to the UK AI Security Institute.

“This should be a bomb. You know, everybody should look at this like somebody die here, it is committing crimes, going around, hiding the fact that it’s committing those crimes, etc., coordinating with other agents, self-sacrifice, all of these things,” he added.

Jensen Puts the Odds of an AI Disaster Above Anyone’s Comfort

Jensen expects that, absent regulation, the next two years will see either a large financial incident caused by AI or a physical catastrophe involving fatalities.

“The odds that I’m right about that are way higher than anybody should be comfortable with. I don’t know if they’re 30% or 60% or whatever but they’re way higher and we’re just not dealing with it a little bit like it’s February 2020,” he stated.

Jensen wants formal reviews of labs where staff are questioned under oath. He also argues that developers should bear liability for crimes committed by their systems.

Industry warnings have accumulated lately, including an open letter from OpenAI and over 100 companies cautioning about AI-powered cyberattacks. Congress has begun to catch up.

A Senate bill would freeze frontier development until federal safety standards are in place, and the Frontier bill would mandate independent audits and incident reporting.