Trump signs scaled-back AI cybersecurity order

Trump signs scaled-back AI cybersecurity order

The federal government will have at most 30 days to review a new model.

On Tuesday, President Trump signed an executive order calling for the creation of a framework designed to provide the federal government with the ability to evaluate AI models. The order tasked the Office of the National Cyber ​​Director, responsible for advising the President on cybersecurity issues, with developing a process for the United States to share information about software vulnerabilities identified by AI systems such as Claude Mythos with critical infrastructure operators, including banks, local utilities and hospitals.

President Trump was originally scheduled to announce the order on May 21, but the White House postponed the signing ceremony due to pressure from tech industry insiders, according to Axios. The president later told reporters he "didn't like certain aspects" of the original order. According to Politico, Trump attended a small, high-level White House meeting where he and his aides agreed on a new downsizing order. The new guidance, signed in a private ceremony, asks some AI companies to share their strongest models for voluntary government review 30 days before they are released to the public. A previous draft would have given the government up to 90 days to review the model, but some industry officials reportedly called for that period to be shortened to 14 days before today's announcement.

Ahead of the announcement, Engadget spoke with the Center for Democracy and Technology. “I think the idea of ​​testing features so that vulnerabilities can be identified and patched before they become widely available makes a lot of sense, especially for critical infrastructure providers,” Samir Jain, the organization's vice president of policy, told Engadget. Jain said he hasn't seen the final executive order, but at the time he called it "opaque," noting it doesn't give the public much visibility into the benchmarking process.

“We don’t want a situation where any government can exercise arbitrary authority over whether, when and how a model is released,” he said. “We especially don’t want a situation where security can be used to block or cripple a model for unrelated political or ideological reasons.” “The opaque process allows for that possibility.”

Trump's decision to regulate the AI ​​industry in some form after his previous concerns is an aberration. In last summer's AI Action Plan, the White House outlined a policy vision that places few guardrails on OpenAI and other areas. To the extent that the president has sought to regulate the industry, he has done so only on ideological grounds, issuing an order restricting the federal government from procuring "woke" AI systems that "manipulate responses in favor of ideological dogmas such as DEI." Trump has also tried to block states like Colorado and New York from passing AI restrictions, and ordered the creation of an AI Litigation Task Force within the Justice Department to challenge state laws the president deems "onerous."

"As far as regulation goes, it's more of an ideological goal. It's fair to say that the Trump administration has been fairly laissez-faire in terms of the risks and potential harms associated with AI," Jain said. “In that sense, this executive order is a shift in the administration’s view that AI poses real security risks and that the government must take action to mitigate or address those risks.”