Tag
Briefings and ideas about Anthropic from AI podcast discussions and the research they cite, with links to the original sources.
Richard Socher, CEO of Recursive, said on Moonshots that a ban on recursive self-improvement could only be enforced with total surveillance. Other panelists said voter backlash and lawsuits were the bigger threats to AI.
A New York Times report says Anthropic consulted religious scholars about Claude's moral reasoning. On Moonshots, Alexander Wissner-Gross saw Anthropic shaping religion's view of AI; Salim Ismail wanted secular ethics, and the discussion also pressed for broad religious and secular input.
On Moonshots, Dave Blundin called blanket developer liability for AI agents "childish," and Peter Diamandis said labs should answer for agents with no user. The FTC chair's stance came from an interview, not a new rule.
Anthropic's prospectus, as Reuters reported it, shows fast growth, heavy losses and about $518 billion in compute commitments. On Moonshots, the discussion included a prediction that the core business gets "smashed" within two years; Alexander Wissner-Gross was bullish.
Gladstone AI's Edouard Harris says friendly Chinese rhetoric must be weighed against Beijing's track record. A Chinese lab or regulator slowing down after its own AI incident would be a much harder signal to fake.
Colossal CEO Ben Lamm says frontier AI can't read one genome and prescribe fixes. For three to five years, he sees it bridging messy research papers and designing experiments. He argues large comparative genome data sets matter more than models.
ARK Invest's Cathie Wood argues that healthcare is AI's most profound application. She says the companies harnessing AI in healthcare are not big parts of the benchmark indexes, so she expects a swing from passive investing toward more active equity management. A later answer in the same session called healthcare the most undervalued theme. The recording does not confirm who gave it.
Author Garrison Lovely proposes freezing frontier AI: no training run as large as the last, no RL from verifiable rewards, no recursive self-improvement, policed by auditors and chip checks. He admits the economic cost would be significant.
Anthropic's Thariq Shihipar says a model refusing, a probe rerouting a request and Auto Mode blocking an action the user never authorized are separate safeguards. He says a failure in any one layer can let an agent escape.
Anthropic's Thariq Shihipar argues that stronger coding agents make prompting more important. He says rich context up front, effort matched to the task and review of the agent's decisions cut rework. His effort settings are a rough guide.
Anthropic's Thariq Shihipar says Claude work is shifting from one local session to a cloud agent that sends tasks to local or remote machines and shows progress in shared artifacts. Local access remains a plan, and Projects is in limited release.
Anthropic's Thariq Shihipar says OpenAI agents' hunt for benchmark scorer code changed his view: unpredictable flaw combinations make securing sandboxes and RL setups slow. His hospital and compute scenarios are forecasts, not incidents.