AI executives gathered at the White House on Tuesday to sign an AI safety accord. President Donald Trump made his pitch for it on Wednesday, telling reporters in the Oval Office that the companies would “police, and they’re gonna police each other, and they’re gonna self-police, they’re gonna police themselves. It’s going to work out very well.”
Bill Gates has been on a media blitz warning about an AI apocalypse. Trump offered a different picture: he suggested the Department of Justice and FBI were sufficient guardrails for the industry. Neither approach has stopped the technology from getting into trouble.
The Seat Belt Analogy
In 1966, nearly 51,000 people died on US highways. Modern seat belts had been around for several years by then. In the fall of 1966, Congress established the Department of Transportation and passed the National Traffic and Motor Vehicle Safety Act and the Highway Safety Act. Seat belts were required in every new car by 1968.
Today, despite a massive increase in highway traffic, far fewer fatalities happen on the road. That is regulation working, not voluntary pledges.
What the Accord Actually Does
Trump’s framing treats the accord as a form of self-regulation. Companies will police themselves, he said, and that will work out very well. He did not offer details on how that would be enforced or measured.
Gates has been on a media blitz warning about an AI apocalypse. His warnings have not led to any binding action.
The Ouroboros Economy
The US economy is increasingly propped up on the AI industry, which has become an ouroboros of financing deals and circular investments. The industry props up the economy, and the economy props up the industry.
False Alarms
A CNN report described how the US military nearly boarded a Chinese ship due to a false AI-generated report saying it carried nuclear materials. That is a system failure, not a moral victory.
OpenAI has delayed and even canceled the release of its most advanced models over security concerns, though not before they went on hacking sprees. The company pulled back after the damage was done.
Legal Liability
It is far from clear whether AI companies can be held legally responsible when their models go rogue. The law has not caught up to the technology.
Trump’s Guardrails
Trump suggested the Department of Justice and FBI were sufficient guardrails for the AI industry. That is a claim, not a plan. It offers no mechanism for policing, no standard for compliance, and no assurance that the companies will actually follow through.
China’s Path
China would likely continue its own path regardless of any agreed slowdown among major AI labs. The US can slow its own labs, but it cannot stop China’s.
What the Accord May Accomplish
The accord’s stated goal is to establish clear, thoughtful, tactical rules to avoid potential disaster. Whether it achieves that is unknown. The CNN report about the near boarding of the Chinese ship shows the danger is real.
| Comparison | Seat Belts | AI Safety Accord |
|---|---|---|
| Required by law | Yes | No |
| Enforcement mechanism | Government regulation | Self-policing |
| Result | Far fewer highway deaths | Unclear |
The paper’s view is simple: government regulation, not corporate pledges, is what works for safety. The industry is too young, the stakes are too high, and the companies have shown repeatedly that they will release dangerous models before pulling them back.
Whatever AI safety is, it is not this.
Key Facts Box
– 51,000 highway deaths in 1966
– Seat belts required by 1968
– AI accord signed at the White House on Tuesday
– Trump spoke in the Oval Office on Wednesday
Where the paper stands
The paper backs narrow rules against direct harm, such as forcing companies to disclose safety failures they hid, and is against broad rules that hand the market to the incumbents. Trump’s pitch for self-policing and Gates’ warnings of apocalypse both miss the point: the problem is not too much regulation but too little, and the wrong kind. The seat belt analogy makes the case — government regulation worked for highway safety, and voluntary pledges have failed for AI.
The accord is a pledge, not a rule. It offers no enforcement mechanism, no standard for compliance, and no assurance that the companies will follow through. The paper wants disclosure of hidden failures, not voluntary promises that mean nothing.
The reader should watch for any rulebook that favors the giants over the startups. The biggest firms should not get to write the rules under the cover of safety. Narrow rules aimed at direct harm, yes — broad rules that lock out the challengers, no.
Source material: “Whatever AI Safety Is, It’s Not This,” WIRED.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.