- Trump signs executive order on AI testing
- Models must undergo cybersecurity review pre‑release
- Major AI firms publicly support initiative
US President Donald Trump signed a new executive order earlier this week, demanding leading AI companies voluntarily submit their flagship models for government cybersecurity testing before deploying them into the market.
This change in philosophy in the Trump administration seems to be fueled by the release of Anthropic’s Mythos Preview, an AI model allegedly so powerful it can surface decades-old software vulnerabilities and develop working exploits. The tool has not yet been released to the public and has instead only been given to a handful of major tech companies, to get a head start on malicious actors. According to Anthropic, the tool was already used to find “thousands” of vulnerabilities, including some deemed critical severity.
Initially, the Trump administration advocated for a more hands-off approach to the tech sector, but now seems set to play a hand in regulating US frontier AI models.
Industry support
According to Reuters, the executive order directs the departments of Treasury, Defense, Commerce, and Homeland Security, as well as other government agencies and officials, to “secure agreements with AI developers to test their models.” The tests would give US agencies a month’s time before the models are released to the market.
Major AI developer companies seem to be on board with this executive order. Google executive Kent Walker allegedly described it as “an important step forward,” and Anthropic said it looked forward to working with the White House. OpenAI CEO Sam Altman said the order “gets the balance right”.
“The US should lead on AI by continuing to develop the very best models, making sure they’re safe, and getting cyber tools into the hands of trusted defenders,” Altman was cited saying.
Reuters also said that voluntary federal testing has been in place “for a few years”, and that major companies, such as OpenAI, Anthropic, have been doing it even during the Biden administration. Last month xAI and Microsoft agreed to do the same thing, although apparently the details “later disappeared from its website”.
Sabeen Malik, VP of Global Government Affairs and Public Policy at Rapid7, commented on the executive order: “The most interesting thing is that both administrations, despite very different philosophies, are converging on the same underlying concern: frontier AI is increasingly being treated as a strategic capability comparable to advanced cyber tools, semiconductors, or dual-use military technologies.”
“The disagreement is no longer over whether frontier AI matters for national security. The disagreement is over whether security is best achieved through regulation and guardrails or voluntary cooperation and competitive dominance. That may end up being the central AI policy fault line for the rest of this decade.”
Via Reuters
The best antivirus for all budgets
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
https://cdn.mos.cms.futurecdn.net/7uv69Evng9VecmGkRPcBuJ-1920-80.jpg
Source link
