AI
AI leaders split on whether the industry will genuinely slow down
Anthropic and OpenAI chiefs propose safety measures to moderate AI development, but skeptics question whether executives are serious about curbing the pace.
TechCrunch — AI
The artificial intelligence industry is caught between competing narratives: some of its most prominent figures are calling for restraint, while others insist the sector must accelerate. Whether this represents a real shift or clever public relations remains unclear.
The safety case for slower progress#
Anthropic's CEO Dario Amodei has launched a push to "pace the frontier," proposing a trio of measures aimed at managing AI-related risks. His suggestions include:
- Independent third-party evaluators stationed inside major AI labs to monitor safety practices
- Coordination among leading AI companies in democratic nations on shared safety standards
- International collaboration on AI development limits
Sam Altman at OpenAI and even Elon Musk have voiced support for some version of this approach. What's striking is how quickly industry consensus seemed to form around these concepts—ideas that had been circulating within AI safety circles for months suddenly gained mainstream backing.
A vagueness problem#
Yet beneath the apparent unity lies a critical gap: the details remain murky. Amodei's proposal outlined principles rather than concrete mechanisms. Critics argue the proposals are light on specifics about both the claimed dangers and what "slowing down" actually means. Both Amodei and Altman tend to frame their position around "pacing" rather than an outright slowdown, a distinction that may matter more than it appears.
The case for skepticism#
On TechCrunch's Equity podcast, observers questioned whether the industry's stated commitment to moderation is genuine. One key concern: these companies may lack structural incentives to truly constrain themselves. The enterprise market—where AI firms now make most of their revenue—tilts against consumer-driven accountability. Corporations won't easily abandon OpenAI for Anthropic's Claude over principle when both are mission-critical. Meanwhile, abundant venture capital allows firms to weather any reputational damage.
Equally problematic is the absence of aggressive government enforcement. Federal regulators have shown little appetite to pursue tech enforcement broadly, let alone specifically in AI.
