How Silicon Valley Killed the White House’s AI Safety Order

Just hours before President Trump was set to sign an executive order requiring government safety reviews of new AI models, he pulled the plug. The reversal came after private calls from Elon Musk, Mark Zuckerberg, and former AI advisor David Sacks, who argued that any oversight would cede America’s lead to China. The order—already voluntary and non-binding—would have created a modest review process for frontier models. But even that was too much. The tech industry’s lobbying machine, backed by hundreds of millions in campaign donations, proved decisive. The decision marks a return to Trump’s hands-off approach and signals that meaningful AI regulation is off the table for the foreseeable future.

The trigger for the White House’s brief flirtation with oversight was Anthropic’s Claude Mythos, a model so adept at finding code vulnerabilities that the company withheld its release. Governments from London to New Delhi panicked over potential attacks on financial systems and critical infrastructure. Yet within weeks, the industry’s influence overwhelmed any security concerns. OpenAI, Google, and Microsoft had already agreed to voluntary, non-binding reviews—a deal that gave the administration cover to back away from stronger action.

With midterms approaching, Silicon Valley’s political spending is only accelerating. Super PACs like Leading the Future, backed by OpenAI’s Greg Brockman, have raised over $125 million to promote anti-regulation candidates. Musk, despite pledging to step back from donations, is again pouring tens of millions into Republican causes. For companies staking trillion-dollar valuations on AI, even a whisper of regulation is a threat too large to ignore.