Despite pleas from industry experts to slow down and regulate the pace of AI, the Trump Administration isn't budging on its approach.
Following a Tuesday meeting with AI industry figureheads, President Donald Trump has rejected the notion that the AI industry needs additional regulation, and that existing laws will be enough to prevent the risks the tech poses.
Instead, Trump and six of AI's biggest names, including Anthropic's Dario Amodei, Google's Sundar Pichai, Meta's Mark Zuckerberg, OpenAI's Greg Brockman, SpaceXAI's Elon Musk and Nvidia's Jensen Huang, signed the White House Accord on Super Intelligence, a code that Trump called "morally binding," promising to self-regulate the risks of development.
The accord states that each company will implement four "layers" of controls and audits, including:
- Implementing robust internal controls to monitor model capability and alignment during training and deployment, with specific areas of risk including cybersecurity, biosecurity and chemical threats.
- Giving an internal team the authority to ensure that all safeguards, controls and monitoring are operating as intended.
- Partnership with external auditors to assess whether these safeguards and monitoring systems are operating properly.
- And creating a designated committee on these companies' boards of directors that oversee the teams running the safeguards and controls and the internal and external teams evaluating them.
The accord states that the signatories would regularly meet to establish standards and improve their safety systems. Additionally, it states that "over time, it may make sense to codify these steps into laws and regulations."
This accord follows Trump rejecting calls for increased regulation and a slowed pace of development, calling the increased AI safety alarmism a "hoax" during an on-stage call with Huang at an event hosted by the "All-In" podcast in mid-September.
Our Deeper View #
While these layers of protection are good in theory, the only thing enforcing this accord is the honor system, rather than actual law and regulation that creates consequences for the risks this tech could present. And though the leaders of these companies have largely been urging for safe development, they are all also the ones consistently pushing the frontier. Real safeguards won't result from a good faith system where companies promise they will try their best to keep the risks reined in. Additionally, these companies holding themselves accountable will only become more difficult as Anthropic and OpenAI approach public market debuts. And while regulation has often struggled to keep up with the pace of technology, in the case of tech that these companies claim could cause catastrophic risk, the best way to get substantial safeguards may be from the real consequences and outside accountability that regulation can provide.