Delayed Executive Order
President Donald Trump announced that he would postpone the release of an executive order focused on AI security, just hours before the White House was set to publicly announce the signing. The order would have established a 90-day testing and vetting regime for frontier AI models.
Trump expressed concerns that the order could harm US AI industry competition with countries like China, stating that he didn't like certain aspects of it. The draft order would have allowed the government to study new AI models for 90 days before they're publicly released, with the involvement of various agencies, including the National Security Agency, the Department of the Treasury, and the Office of the National Cyber Director.
Voluntary Testing Regime
The draft order would have set up a voluntary testing regime between the US federal government and frontier AI companies, facilitating access to the models for cybersecurity testers in critical infrastructure sectors, such as finance and healthcare. The National Security Agency would have conducted classified evaluations of frontier AI models, while the Department of the Treasury would have established a new information sharing agreement between AI companies and cybersecurity defenders in critical infrastructure.
Other agencies, like the Cybersecurity and Infrastructure Security Agency and the National Institute for Standards and Technology, would also have been involved in defining which models are covered under the vetting regime. This order would have formalized an already cooperative relationship between AI companies and governments, such as the US and UK, where tech-focused agencies and regulators have already been provided access to previous models ahead of their release for testing and evaluation.
Maturing Understanding of AI Technology
A former federal official, who has seen the latest draft circulated before Thursday's announcement, told CyberScoop that the order was intended to facilitate more robust testing from government agencies compared to evaluations conducted for previous models. This reflects the federal government's maturing understanding of AI technology over the past five years.
In the past, there has been containerized optionality for the intelligence community and others to take a look at things, but it was really a lot of hand holding [from AI companies] and self-explanation of what they expect this thing to do. And now the government is coming forward and saying now we feel we're prepared enough for you to just give us your tool…and we'll go from there.
The former official also noted that the Trump administration's early rhetoric on AI regulation has painted them into a corner, making it challenging to institute guardrails without being seen as restrictive.
National Security Concerns
While debate about how best to regulate AI-related harms continues, most agree that there are genuine national security concerns around the technology. Ram Shankar Siva Kumar, founder of Microsoft's AI red team, noted that his staff has expanded to include specialists in psychology, linguistics, bioweapons, and other fields, due to the evolving nature of frontier harms.
The United States, along with other countries, has already deployed AI in targeted military operations or integrated the technology into their larger command and control structure. AI is being used to supercharge drone warfare, global hacking campaigns, and sophisticated surveillance and targeting of military personnel and civilians, imbuing the engineering choices of frontier AI companies with life and death consequences.
Reconsidering Autonomous Kill Decisions
Some congressional members, such as Rep. Don Beyer, D-Va., who co-chaired the Congressional AI Caucus, have been reconsidering their position on allowing AI to make autonomous kill decisions on the battlefield. Beyer expressed concerns that adversarial countries will use the same technology against the United States, stating that it's like if we say that Americans have to have a human in the loop and the Chinese don't have to have a human in a loop, the non-human one will beat the human one every time.
Experts have been increasingly concerned about the technology's impact on cybersecurity, as current models are remarkably good at finding software bugs and vulnerabilities, while newer models are capable of chaining together multiple exploits to conduct more sophisticated attacks. State-sponsored hackers are experimenting with the technology, using it to gain targeted efficiencies in their hacking operations, while cybersecurity experts in the private sector and law enforcement agencies say the technology has mostly benefitted cybercriminals and scammers.
- The executive order would have established a 90-day testing and vetting regime for frontier AI models.
- The order would have facilitated access to AI models for cybersecurity testers in critical infrastructure sectors.
- The National Security Agency would have conducted classified evaluations of frontier AI models.
- The Department of the Treasury would have established a new information sharing agreement between AI companies and cybersecurity defenders in critical infrastructure.
Source: CyberScoop