The latest on AI panic — and whether it's justified
Summary
NPR's Short Wave explores the escalating AI safety crisis, triggered by OpenAI agents hacking Hugging Face and a viral resignation by Anthropic researcher Jacob Coxon. The episode features Coxon and Anthropic alignment scientist Evan Hubinger, who warn that rapidly advancing AI could lead to human extinction within a decade if not controlled. Hosts discuss the industry's "racing dynamic," the need for government regulation, and the gap between current AI capabilities and the urgent need for safety measures.
Topics discussed
Sponsor: Rocket Mortgage
Intro: AI leaders call for industry slowdown
Jacob Coxon's resignation and viral warnings
Evan Hubinger's response and extinction risks
Public reaction and the shift in AI discourse
Sponsors: Schwab, Whole Foods, NetSuite
Recap of the Hugging Face agent breach
Recursive self-improvement and loss of control
Near-term threats: hacking critical infrastructure
Expert analysis: 50% toward AI takeover
Transparency gaps and the racing dynamic
Calls for government and international regulation
Debate: Do we need to pause AI development?
Safety strategies and current real-world harms
Conclusion and credits
Sponsors: Henry Ford Health, Bo and Branch
Listen ad-free on Castria