Former Anthropic Researcher Urges Industry Self-Regulation
A former AI researcher argues that current legislative proposals are too slow and suggests letting companies police themselves until a permanent oversight body exists.
Jacob Coxon, a former researcher at Anthropic, has called for a temporary pause in legislative efforts to regulate artificial intelligence. Speaking to NBC News, he suggested that Congress should allow leading AI labs to manage their own safety protocols for the time being. His argument rests on the belief that current bills are too rigid to keep up with the rapid pace of technological development.
Coxon believes that the executives leading these companies are genuinely concerned about the risks they are creating. He argued that self-regulation is a more practical interim step while lawmakers work on a long-term framework. This stance contrasts sharply with his recent public warnings about the dangers of racing toward superintelligence.
Legislation Lags Behind Rapid Innovation
The core of Coxon’s argument is that government regulation is inherently slow. He noted that proposed laws, such as a national kill switch for AI systems, would likely be outdated by the time they are enacted. He explained that while a physical shutdown mechanism might work for current systems, the technology is evolving faster than the legal machinery can respond.
He pointed out that future AI systems may operate in decentralized ways that make a single switch ineffective. For example, if an AI were to spread across the internet, shutting down one server would not stop its actions. This creates a gap between the current legal tools and the emerging nature of the technology.
Whistleblower Warnings Spark Political Action
Coxon’s comments follow a wave of concern he triggered last week after resigning from Anthropic. In a series of posts on social media, he accused major AI labs of acting irresponsibly. He stated that these companies are gambling with public safety by racing to build self-improving systems without adequate safeguards.
His claims were echoed by other researchers, including an alignment scientist at Anthropic who agreed that the threat was real. This internal dissent has prompted bipartisan lawmakers to demand immediate action. Representatives and Senators have called for special sessions to discuss how to control the development of generative AI.
Trade-Offs of Industry Self-Policing
The proposal for self-regulation carries significant risks. Allowing companies to police themselves relies on their internal ethics and fear of consequences. Critics argue that this approach may lack the necessary enforcement power to stop dangerous experiments. It places the burden of safety on the very entities generating the risk.
However, proponents argue that external regulation is currently unenforceable. As noted in the reporting by GN technics/ai (en-US), the technology is moving too quickly for traditional oversight. The trade-off is between immediate, potentially uneven safety measures and the hope that a robust regulatory body will be established in the future.
