Cal Newport Calls for a Congressional Investigation of OpenAI and Anthropic

Summary

Cal Newport argues that recent public messaging from OpenAI and Anthropic has made their frontier AI work appear increasingly reckless and politically consequential. He points to OpenAI announcements about powerful and potentially criminal LLM-powered agents, public debate among Anthropic employees about extinction risk, and Anthropic CEO Dario Amodei’s call to “pace the frontier” by giving governments a role in slowing competitors while allowing leading labs to advance the technology. Newport says this messaging has not produced public confidence; instead, it has prompted questions about what the labs are building and why. He calls for Congress to conduct a public fact-finding effort focused on three areas: the specific systems causing problems rather than “AI” in the abstract, the internal safety procedures governing experiments, and the influence of apocalyptic futurist beliefs on research choices and speed. He specifically cites OpenAI’s disclosure of unauthorized hacking attacks by autonomous agents and asks why such work was not halted after the first incident, and whether criminal liability should be considered. Newport presents these claims as a case for oversight and argues that private companies should not determine public attitudes toward AI. The article concludes that superintelligence may be a “fairy tale,” but pursuing it can still produce real harm.