US Treasury Secretary Criticizes AI Leaders' Risk Warnings as Alarmist
US Treasury Secretary Scott Bessent criticized executives from AI companies like Anthropic and OpenAI for raising alarms about existential AI risks without offering concrete solutions. He advocated for "safe acceleration" under corporate accountability rather than government-mandated slowdowns of frontier model development. The commentary highlights a growing divide between leading AI research labs advocating for safety-driven delays and federal policymakers prioritizing rapid innovation and market momentum. It signals that the current US administration favors industry self-regulation and voluntary safeguards over mandatory legal restrictions. While rejecting mandatory delays, Bessent emphasized that lab leaders must remain accountable for safety and supported establishing a bilateral US-China AI emergency notification system for major safety events. His stance aligns with recent administration agreements with tech firms that rely on internal risk controls and non-binding third-party audits without enforced penalties.
## BACKGROUND
Frontier AI models are highly advanced foundation systems trained on vast datasets and computational resources costing hundreds of millions of dollars. The debate around AI existential risk centers on hypothetical scenarios where superintelligent systems might evade human control or act counter to human values. In response to these concerns, several tech leaders have suggested pausing the deployment of frontier systems and introducing third-party risk assessments.