Amongst a slew of executive decisions on Induction Day, President Donald Trump signed one revoking the Biden-era Executive Order on AI Safety

By Justine Dukmedjian.

On January 20th, 2025, President Donald Trump revoked former President J. R. Biden 2023 [2] [3] executive order on AI safety regarding consumers, workers, and national security [1]. Even though this revoking came with the slew of executive orders the U. S. president has signed since Induction Day [4], it is worth taking notice of it. This decision is linked to the Republican Party’s vow to get rid of anything that could hinder freely made “AI development.”

Biden’s executive order was written and signed at the end of 2023. This was when AI – especially generative ones – was starting to unleash its potential to a broad audience, nearly a year after ChatGPT’s public release. Firms like Palantir Technologies and agencies such as the CIA or the FBI have been using AI for years. Yet, the introduction to the public of GPT and GPT-like technologies, as well as their exponential progress, brought AI use to a whole new level. This is why former President Biden and his administration worked on a legal framework meant to ensure AI’s safe development. The Biden administration had the safety of all parties involved in mind – such as protecting people from biases and discrimination or larger security concerns (national security implications, for example)[5]. So, even though it could have been flawed in its shape, the logic behind the making of this executive order seems sound: safeguarding the American people from the potential dangers of the rising omnipresence of AI technology.

Even Elon Musk, President Trump’s new friend and ally, seemed to agree with the risks inherent to the quick evolution of AI since November 2022. In March 2023, he called for a memorandum on AI development alongside other tech experts [6] [7]. Which makes President Trump’s urge to sign this executive order – he did it as soon as he got into office – all the more surprising, if not puzzling. Indeed, one of the first things President Trump did after taking office was to revoke an order precisely aimed at addressing the concerns raised by Musk and his peers.

So, it begs the question: why do that? Whose idea was this? Musk’s? Another tech mogul’s – like Peter Thiel, whose Palantir pioneered AI use? Donald Trump’s himself? Or did it come from Donald Trump’s newfound supporters who were interested in the matter, like Bezos or Zuckerberg? It’s hard to tell. The only sure thing is that revoking this executive order without drafting a new one addressing the same issues might be putting America at risk. Especially since AI developers are no longer required by law to conduct safety tests. Nor are they mandated to communicate any voluntary safety test results with the federal authorities – even if they might put in danger the security of the nations; their inhabitants; patients; etc.

Even if President Trump’s decision seems inscribed in his broader philosophy of innovation and freedom, he might be playing with fire with this one. Especially in an era where cyberattacks from traditional auto-proclaimed “enemies” of the U.S.A. – like China, Russia, or Iran – are multiplying [8] and the perspective of a global conflict (where AI safety could be paramount to lead to victory) looms.

Edited by Justine Peries.