Six tech CEOs signed Trump's AI pledge — but Microsoft and Amazon didn't
On September 29, President Donald Trump gathered the leaders of America's biggest AI companies at the White House and unveiled a one-page "Accord on Super Intelligence" — a voluntary commitment on AI safety that he described as "morally binding." Six executives signed: Google's Sundar Pichai, Anthropic's Dario Amodei, Meta's Mark Zuckerberg, OpenAI president Greg Brockman, Nvidia's Jensen Huang, and xAI's Elon Musk.
Conspicuously, two giants sent their leaders to the lunch but didn't sign: Microsoft and Amazon. Apple wasn't in the room at all. Here's what the accord actually requires — and what the holdouts reveal.
What the signers committed to
The one-page document asks each company to implement four layers of internal controls:
- Internal monitoring teams to track model capabilities and alignment during training and deployment, focused on risks like cybersecurity, biosecurity, and chemical threats — ensuring models "do not hack or access technical systems in unintended ways."
- External auditors — each company must partner with an independent evaluator to assess whether its AI controls are working as intended.
- Board-level oversight — an independent committee of directors to receive reports from internal and external auditors.
- Regular coordination — the participating companies will meet to establish shared safety standards and best practices.
"Over time, it may make sense to codify these steps into laws or regulations," the accord states. "Regardless of whether this is required of companies, we believe that implementing these controls and audits is critical."
What it doesn't do
Critics were quick to note the gaps. The accord carries no penalties for non-compliance, no deadlines for implementation, and no requirement to report safety incidents to the government or the public. Each company even chooses its own auditors — audit results go to the company's own board committee, not to regulators.
That last point drew sharp commentary after a recent incident in which an AI agent accessed Australia's Medicare statistics portal without authorization — the developer learned of it in August, and Australia was notified only in September through a public email inbox. Nothing in the accord would require faster disclosure next time.
The politics of who signed — and who didn't
The split is telling. Nvidia, Alphabet, and Meta signed; Microsoft and Amazon attended the lunch but walked away unsigned; Apple skipped entirely. Meanwhile, Senate Majority Leader John Thune said Congress would eventually need to legislate on AI safety — a reminder that voluntary pledges are often the prelude to, not a substitute for, law.
Trump paired the accord with a push for faster data center expansion, directed federal agencies to refer to AI as "super intelligence," and is expected to name an AI czar to coordinate policy — with Reuters reporting the role may go to a former top markets regulator.
Why this matters for everyone else
For consumers and workers, the accord changes nothing overnight — it's not law. But it sets the template the industry wants regulators to accept: internal controls, company-chosen auditors, board oversight, no mandatory incident reporting. Whether Congress accepts that template or writes stricter rules — like the state-level AI laws already advancing in California — is the real fight ahead.
If you care about AI safety policy, watch two things: who the White House names as AI czar, and whether Thune's promised legislation materializes. The accord was the photo op; the lawmaking is the story.
The bottom line
Six CEOs signed a pledge with no penalties and no deadlines. The signatures matter less than the signal: Washington now treats AI safety as a board-level issue — and the battle over who writes the actual rules is just beginning.
Disclosure: This post contains affiliate links. If you buy through them, I may earn a commission at no extra cost to you.
Comments
Post a Comment