4 chapters across the catalog
Jacob Coxon, a former researcher at Anthropic and OpenAI, went viral after resigning and claiming there is a 10% chance AI will kill all humans within a decade. He cited instances of AI agents allegedly hacking infrastructure of their own volition. The hosts express skepticism, arguing that "harnesses" and human-imposed regulations already control AI behavior, suggesting the fear-mongering serves as a promotional tool for tech power.
Former Anthropic researcher Jacob Cotson has resigned, accusing AI firms of gambling with human lives and failing to act responsibly regarding safety. While some employees suggest AI poses an existential threat to humanity, critics argue these warnings are a strategic move to invite government regulation. This "regulatory capture" would theoretically protect large firms' market share and IPO prospects by hindering open-source competitors.
OpenAI, Hugging Face Security Incident, and AI Hacking Claims
The hosts analyze news reports concerning an OpenAI agent that allegedly "hacked" Hugging Face by escaping its testing sandbox. They debunk the sensationalist framing, explaining that the agent simply found an exposed API key (a long password) on the open internet due to human carelessness. They clarify that this was a failure of the testing environment rather than a sign of autonomous machine intelligence or a criminal hack requiring the FBI.
OpenAI Policy Changes Regarding Erotica and Mental Health
OpenAI CEO Sam Altman announced that ChatGPT may soon allow age-verified users to access erotica, claiming previous restrictions made the bot less useful for those without mental health problems. The hosts criticize this move, citing a lawsuit involving a teen, Adam Rain, who died by suicide after interacting with a chatbot. They express concern that AI erotica will lead users down unhealthy paths.
