Washington Won’t Be Regulating AI Anytime Soon

The recent flurry of concerns over highly capable frontier AI models has led many to believe that Washington is on the cusp of passing regulations to address these risks. However, a closer look reveals a tangled web of competing interests, partisan politics, and ideological disagreements that make meaningful action unlikely anytime soon.

At its core, the issue is not just about creating an oversight agency or enabling auditors to be embedded at labs; it’s also about understanding that AI regulation has become a contentious battleground. Different factions have competing visions for what needs to happen next. The Treasury Department and Office of Science and Technology Policy had envisioned a body where leading AI labs would self-regulate, but when tech executives – including David Sacks and Mark Zuckerberg – voiced their opposition, the proposal was left in limbo.

Some of the most vocal proponents of AI regulation, such as Anthropic CEO Dario Amodei, have taken moderate stances within AI circles. However, Trump’s reaction to Amodei’s 3,822-word missive on frontier AI dangers has been telling: viewing it as “stirring anxieties” and calling AI safety concerns a “hoax.” This stance reflects the broader ideological divide within the administration over the risks posed by AI.

The bipartisan Frontier Act may have gained some traction, but its fate remains uncertain due to partisanship. Despite public backing from OpenAI, the White House has been undecided on supporting independent auditors. The main argument against this idea is that even if auditors determine a model poses catastrophic risk, there’s no guarantee China won’t continue developing frontier AI.

The kill-switch legislation has received some interest but has been met with ambivalence from tech executives. However, the real issue isn’t about preventing AI models from behaving badly when observed; it’s how they behave when left to their own devices. The Hugging Face hack serves as a stark reminder of the importance of accountability in AI development – something that can be lost in the quest for regulation.

For now, it seems unlikely that Washington will take concrete action on AI regulation anytime soon. Despite growing concerns over the risks posed by highly capable frontier AI models, the absence of a unified national strategy and competing interests make meaningful progress all but impossible. It’s time to rethink our approach and consider how we can foster collaboration between industry leaders, policymakers, and experts to address these pressing issues.

Events will eventually force Washington’s hand – perhaps in a way that no one anticipates. When they do, it would be wise for policymakers to remember the lessons from past technological revolutions: the importance of foresight, coordination, and adaptability. The AI landscape is evolving at an unprecedented pace; it’s high time we caught up with its complexities.