What a time to be live. The main thing happening today is the viral post questioning who really funds AI safety. Live from 12–3 BST. Here's your Order of Play.
12:50 - Binji Pande, Founding Member at Ethlabs
13:00 - Matthew Wilson, Co-Founder & CEO at Jack & Jill
13:15 - Lukas Saari, Co-Founder & CEO at Tandem Health
13:30 - Niccolò de Masi, Chairman & CEO at IonQ
13:50 - Kaspar Klemm, Co-Founder at Cepter
14:00 - Eric Queathem, Founder & CEO at Velocity
14:15 - Giovanni Alberto Falcione, Co-Founder & CTO at Exein
14:30 - Stefan Gerbes, Co-Founder & CEO at Nomos
What’s Happening Today
A viral audit this week claims the AI safety world is one big conflict of interest.
Kevin Bass, a blogger and vocal critic of Anthropic, published a financial audit arguing that the company has built what he calls a "regulatory capture machine." His argument is that METR, the independent evaluator Dario Amodei proposed to assess AI models, is financially connected to a wider network of organisations tied to Anthropic.
At the centre is Dustin Moskovitz, an early Anthropic investor whose donated stake was estimated at around $500 million and could now be worth more than $7.7 billion. Bass points to his funding, which helps organisations including METR and Redwood, as part of a much wider network of groups working on AI safety and risk. He argues that some of those same organisations are also behind the growing warnings about AI doom. He's now calling for a Congressional investigation.
The thread traces a web of connections between Anthropic investors, AI safety organisations and the groups funding them. METR lists people from Jane Street among its donors, while Anthropic's investor base includes Jane Street. Bass also points to groups such as the Tarbell Center, which promotes warnings about AI risks.
It comes at a time when Anthropic's approach to AI safety and regulation is facing growing scrutiny, while its relationships with outside evaluators and safety organisations are attracting more attention. Bass is calling for Congress to investigate the relationships behind the network.
We're getting into all of it on today's show at 12 BST.
Daily Op-Ed - Anthropic Doom Timeline
This morning I thought giving a timeline of what’s been happening over the past 7 days was needed because the timeline has been in turmoil, debates are flying everywhere and now we have a tin foil hat.
Timeline so far:
September 8: Jacob Coxon (worked on pre training at OAI and Anth) resigned after about 4 months and did so by posting a viral X thread. He said both labs were “racing straight to self-improving super intelligence and gambling with our lives,”. This obviously broke the timeline and achieved escape velocity incredibly quickly. For example, yesterday Ronan was talking about the wedding he attending on the weekend, saying he was at the brunt of questions about AI killing all humans.
September 9–11: Anthropic Alignment Lead Evan Hubinger publicly agreed that researchers “earnestly believe AI could kill all humans” and put his personal odds above 10% within a decade. Coxon gave interviews with trad media, calling for regulation while describing RSI as potentially imminent. On the timeline, this framed him as a credible insider whistleblower - debate ensued.
September 12–13: Debate polarised supporters treated the warnings as overdue honesty from the labs. A lot of the timeline was calling it “fear marketing”.
September 14: Kevin Bass’s thread then tied the episode to a larger alleged “Anthropic Network” funding loop and explicitly presented Coxon’s resignation as the public face of that self-reinforcing cycle. Bass goes further than the plant theory alleging Coxon as an insider product of a tightly funded safety/doom ecosystem whose incentives point the same direction as Anthropic’s regulatory-capture interest.
The thread maps a network of interconnected organisations (including Coefficient Giving, Redwood Research, FAR AI, Tarbell Center, and Longview Philanthropy) that promote AI risk narratives while positioning the same groups to assess and regulate Anthropic's models.
What that means: The structural incentives lock the ecosystem into amplifying "AI doom" messaging for growth and regulatory advantage, potentially slowing US AI progress relative to China. He concluded the post by calling for Congressional investigation. We’ll run through the post on today’s show.
- Luke
etn. Insight
Ed Leon Klinger, co-founder and CEO of Flock, came on the show and echoed Dario Amodei’s warnings about AI agents.
“I imagine there’s going to be a cascade of news stories over the next three months,” he said, as agent swarms start causing trouble. Still, slowing down while rivals race ahead isn’t an option. Neither is charging toward superintelligence and brushing aside the risks. “We have to get the balance right.”
Watch the full clip below.
Continental Breakfast
etn. is made possible by Airwallex, base, ElevenLabs, Framer, HEX, BVNK, Valarian, Metaview, Chargebee and Polymarket.