- The incoming Trump admin reversed Biden's 2023 AI safety executive order as a first order of business
- Analyst Jack Gold said that this will toss AI responsibility to the states
- He thinks it will also mean that the EU sets the AI rules for the next few years
Newly appointed President Donald Trump used one of his first pen strokes to reverse former President Joe Biden’s executive order on artificial intelligence (AI) safety on Monday, effectively removing federal oversight of the technology. Just a day later, Trump talked up a new four-year, $500 billion investment in AI infrastructure from Open AI, Oracle and Softbank dubbed "Stargate." But analyst Jack Gold warned this all gas, no brakes approach could well prove dangerous.
Amid a flurry of executive orders the new admin issued on day one, Trump halted enactment of a 2023 Biden order that made leading AI companies – such as Google, IBM and Open AI – share AI safety test results and other crucial information with the federal government. The Biden AI plan had also called for the creation of a U.S. AI Safety Institute to create voluntary standards for the use of new technology.
Trump has already made venture capitalist David Sacks his AI czar. Under Sacks’ watch, the government is likely to take a much more non-interventionist approach to AI regulation, leaving any heavy lifting on the technology to the states. California already passed initial AI regulations in October 2024.
“Basically they are removing restrictions and making it a market choice issue,” commented principal analyst Jack Gold of J.Gold Associates. “It's unlikely that the Feds will enact anything here for the foreseeable future, unless the legislature (also controlled by the Republicans) decide to do something."
Gold thinks that the new admin’s lassiez-faire attitude could be dangerous. “Leaving this to the companies to self-regulate is dangerous in my opinion. What’s their incentive to make things safe and secure, with market pressures wanting them to pursue rapid updates and product releases?“ Gold asked.
Gold's question is especially relevant in the face of skyrocketing investment in AI infrastructure in the U.S. In addition to the $500 billion Stargate initiative mentioned above, Microsoft has also outlined plans to spend $80 billion on AI data center expansions in its fiscal 2025 alone. Likewise, Amazon is expected to invest tens of billions of dollars in data center expansions, while smaller players like Vultr and CoreWeave are also plotting rapid growth.
But Gold noted “AI use can cause some real world damages if not done right." Scientific American has already named some real world AI problems, such as false arrests, unwarranted surveillance and deep fake pornography.
The analyst doesn’t think that the Republican-led Congress will enact any AI safety regulations under Trump and doesn’t believe the states will be able to wield much influence either. “Given the issue of where the AI models actually run, which is usually in data centers in the cloud and scattered all over even when states, like CA, do regs, it usually takes years to be put in place,” he said.
Rather, Gold thinks that the European Union (EU) will likely hold the whip hand on AI regulations over the next few years. “The Europeans are taking a strong stand on this and we could see companies that work internationally (virtually all the AI companies) have to conform to the EU and just offer the same for U.S. customers, especially internationals,” he concluded.