Two US senators, Josh Hawley and Richard Blumenthal, are taking a bold step towards regulating artificial intelligence. Their new bill aims to create a federal program at the Department of Energy to evaluate the risks of advanced AI systems.

The Artificial Intelligence Risk Evaluation Act would require developers to submit their models for review before deployment. This is a significant departure from the usual ‘move fast and break things’ approach in Silicon Valley. The bill also echoes a recent landmark AI law passed in California, which focused on consumer safety and transparency.

The Growing Concerns Around AI

While some might see this as an overreach of government control, the concerns around AI are very real. Rogue systems, security breaches, and even weaponization by adversaries are all potential risks that need to be addressed. By creating a program to gather data on these potential disasters, the government can better understand the scope of the problem.

Hawley and Blumenthal’s bipartisan effort is a sign that both parties are finally taking AI seriously. Their previous proposal to shield content creators from AI-generated replicas of their work shows they see AI as a double-edged sword – capable of creativity and chaos in equal measure.

The Debate Over Regulation

However, not everyone is on board with the bill. The White House has expressed concerns that over-regulation could dampen innovation and put the US behind in its AI race with China. This tug-of-war between safety and speed is a familiar one in the tech world.

The recent Snapdragon Summit showcased the rapid progress being made in AI-driven technologies, from laptops to ‘agentic AI.’ Policymakers are scrambling to catch up, but it’s refreshing to see lawmakers trying to address these questions before catastrophe strikes.

A Crucial Step Towards Responsible AI

Bills like this one won’t fix everything, and they might even slow down some flashy rollouts. But can we really afford another ‘social media moment’ where we realize the risks only after the damage is done? I’d argue that common-sense oversight is less about stifling progress and more about ensuring that progress doesn’t come back to bite us.

The outcome of this bill is uncertain, but one thing is clear: AI has officially moved from tech blogs to the Senate floor, and it’s not going back. Whether or not this bill gains traction, it’s a crucial step towards responsible AI development – and a necessary one at that.

Inspired by

• https://ai2people.com/senators-push-new-ai-risk-bill-could-this-be-the-first-real-guardrail-on-artificial-intelligence/