Blueprint
Regulation, liability and the doom narrative
Sam's argument that the labs' safety messaging is regulatory capture, his proposal that labs carry large liability cover, the hosts' security predictions, and Brit's counterpoint that the safety concerns are real.
Voice and stakes
Voice: mostly one voice (Sam, ~45 % of attributed items). Brit’s view that the safety concerns are real is the main counterpoint; Dave and Jess mostly relay, and Jess adds reporting. This page records host opinion on political and regulatory questions, not a position of the wiki.
Stakes. Slow Ventures, Sam’s firm, is an AI investor, and the corpus flags this on the liability proposal (Sam · The Real Reason Elon Wants Sp (Oct 2026) ▶ 10:26). Dave co-founded a home-security company, Sauron (Dave · Apple’s iPhone Sock, Waymo’s B (Nov 2025) ▶ 26:21), and Sam’s firm had an Instagram account hijacked (Sam · Why SpaceX’s $1.75T IPO Makes (Jun 2026) ▶ 25:07). Jess’s outlet, The Information, broke the Dario memo story the hosts discuss (Jess · Fire Sam Altman, The End of So (Mar 2026) ▶ 03:01). Jess says none of them are Washington experts (Jess · Anthropic Wants to Slow Down. (Sep 2026) ▶ 44:04).
The argument in brief
- Sam argues the labs’ scary narrative is also how they raise money and gain policy leverage (Sep 2026) (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 13:57).
- He argues that when labs cannot win on economics they switch to politics, the classic move of changing the game (Jul 2026) (Sam · Open Source AI Wins. Now Labs (Jul 2026) ▶ 22:17).
- Sam argues no regulatory roadmap makes sense, and unilateral national action is irrelevant in a connected world (Sam · Open Source AI Wins. Now Labs (Jul 2026) ▶ 25:32).
- Sam proposes that labs carry roughly $100 billion of private insurance or fund a contingency pool, because they centralise profits and radiate risk (Oct 2026) (Sam · The Real Reason Elon Wants Sp (Oct 2026) ▶ 10:53).
- Sam predicts regional banks will be hacked, and that the labs become valuable mainly as defence companies (Sam · The Real Reason Elon Wants Sp (Oct 2026) ▶ 03:12) (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 18:52).
- Brit says the safety concerns are real, not just marketing, but the US cannot slow down because of China (Brit · Anthropic Wants to Slow Down. (Sep 2026) ▶ 08:34).
- Dave offers two readings of the Anthropic media tour: a planned play for capture, or a real sign of power in the newest models (Dave · iPhone Duo, Meta Muse & Why AI (Sep 2026) ▶ 32:29).
The narrative as leverage
Sam says the lead labs use the scare narrative as marketing for advantage while being in a commoditising business (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 10:11). His words: (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 13:57) In April he says it is not a branding problem, since telling Washington and investors that AI is dangerous was the labs’ own money-raising strategy (Sam · What Tim Cook’s “Executive Cha (Apr 2026) ▶ 06:46). He says AI wars are now politics, with national security and child safety the two arguments that end debates (Jul 2026) (Sam · Open Source AI Wins. Now Labs (Jul 2026) ▶ 19:29), and that a commodity utility can respond with regulatory capture or another route (Sam · Elon Musk vs OpenAI Trial, Goo (May 2026) ▶ 26:31). He admits he is depressed by his own devil’s-advocate position and offers no advice (Sam · Open Source AI Wins. Now Labs (Jul 2026) ▶ 19:26).
Others partly agree. An unnamed host says the hatred of AI comes partly from labs leaning into scary marketing (one of the hosts · Google I/O Just Killed SEO? Ag (May 2026) ▶ 32:12). Dave says the doomer story uses the same fear trick as an old institution and may be the right narrative technology to win (May 2026) (Dave · Elon Musk vs OpenAI Trial, Goo (May 2026) ▶ 48:41); he also says scaring people may not work, but there may be no better way to start the conversation (Dave · iPhone Duo, Meta Muse & Why AI (Sep 2026) ▶ 35:22). An unattributed host calls the slowdown letters emotional noise to be ignored (one of the hosts · Meta’s AI Campaign, While Riva (Jul 2026) ▶ 51:15).
Washington reporting is relayed. Jess relays that a well-connected adviser summed up the DC mood about Anthropic in one word, money (Jess · Anthropic Wants to Slow Down. (Sep 2026) ▶ 15:17), and that tech people are realising they need Democrats on their side (Jess · Anthropic Wants to Slow Down. (Sep 2026) ▶ 42:52). Sam predicts policy staff will be paid about ten times more (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 14:38); that figure is a prediction, not data. Jess says the government process around the Fable clearance looks politicised (Jul 2026) (Jess · Every CTO Is Quietly Switching (Jul 2026) ▶ 37:40); Sam says it is trust-destroying but has marketing value, and he cannot tell whether it is strategy or intuition (Sam · Every CTO Is Quietly Switching (Jul 2026) ▶ 38:52) (Sam · Every CTO Is Quietly Switching (Jul 2026) ▶ 39:30).
Dave says he has heard for months that Anthropic tried to convince the White House to make open source illegal (Dave · iPhone Duo, Meta Muse & Why AI (Sep 2026) ▶ 32:54). This is hearsay. Dave, relaying Bill Gurley, calls the idea nonsense for a free market (Dave · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 28:45), and Jess hears buzz about open-source action because the labs fear Chinese models (Jess · Open Source AI Wins. Now Labs (Jul 2026) ▶ 12:23).
What the hosts say regulation should and should not do
- States. In November 2025 Sam says fifty state frameworks is “a great way to lose” (Sam · Google Just Won AI, TikTok’s “ (Nov 2025) ▶ 45:58) and that states are subscale (Sam · Google Just Won AI, TikTok’s “ (Nov 2025) ▶ 43:10). Jess pushes back that states regulate other things and recounts, with hedging, California’s vetoed bill (Jess · Google Just Won AI, TikTok’s “ (Nov 2025) ▶ 41:59) (Jess · Google Just Won AI, TikTok’s “ (Nov 2025) ▶ 42:04).
- Government equity. Sam thinks the government should not take stakes in AI companies (Sam · SpaceX Hit $2.5 Trillion. Now (Jun 2026) ▶ 37:34); Jess dislikes tying software access to nationality (Jess · SpaceX Hit $2.5 Trillion. Now (Jun 2026) ▶ 35:15).
- Federal posture. Jess reports the White House briefed AI companies on plans to pre-review models (May 2026) and does not know how it ends (Jess · Google I/O Just Killed SEO? Ag (May 2026) ▶ 44:10) (Jess · Google I/O Just Killed SEO? Ag (May 2026) ▶ 44:55). Amir Efrati says the government is heading to early access to models rather than vetting (Amir Efrati · Amir Efrati (May 2026) ▶ 09:51).
- Child safety. Sam calls Meta’s settlement theatre and wants one national rule for every platform (Sam · Meta’s $17B Reckoning & Why Pa (Aug 2026) ▶ 28:59) (Sam · Meta’s $17B Reckoning & Why Pa (Aug 2026) ▶ 17:25); Dave calls it a nothingburger (Dave · Meta’s $17B Reckoning & Why Pa (Aug 2026) ▶ 08:38); Jess backs limits though she doubts they work (Jess · Meta’s $17B Reckoning & Why Pa (Aug 2026) ▶ 13:10).
Liability
Sam’s frame is that technology centralises value and radiates risk, a line he attributes to Jaron Lanier in one episode (Sam · Matt Mazzeo (Sep 2026) ▶ 18:57) and to a Kevin Kelly-style idea he cannot source in another (Sam · Why OpenAI Can’t Win The Decen (Feb 2026) ▶ 38:14). Applied to the labs: (Sam · Matt Mazzeo (Sep 2026) ▶ 19:30) His principle is that if you create the risk you should carry the balance sheet (Sam · The Real Reason Elon Wants Sp (Oct 2026) ▶ 10:46), and he predicts Katrina-sized lab-leak damage (Sam · Silicon Valley Is Great at Tec (Oct 2026) ▶ 20:32). He says the sensible framework will never happen (Sam · Silicon Valley Is Great at Tec (Oct 2026) ▶ 18:55).
Agents come in too: who is liable for an agent decides whom it aligns to, and the law will lag the building (Sam · Matt Mazzeo (Sep 2026) ▶ 42:46) (Sam · Matt Mazzeo (Sep 2026) ▶ 42:27). Jess reports Google Workspace terms make her responsible for what her agents do (Jess · The Real Reason Elon Wants Sp (Oct 2026) ▶ 11:50). Sam’s Disney story makes the point that public, suable companies are trustworthy because liability has a price (Sam · The Internet Is Fighting Back (Oct 2026) ▶ 14:01). In tension with all this, he says it is unclear anyone could get money by suing OpenAI or Anthropic (Sam · The Internet Is Fighting Back (Oct 2026) ▶ 14:10).
Security predictions
Sam says you do not need superintelligence for chaos (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 09:19) and that all regional banks will be hacked (Sam · The Real Reason Elon Wants Sp (Oct 2026) ▶ 03:12). Dave relays that new models can reverse-engineer any binary, so software is effectively open source (May 2026, secondhand) (Dave · Amir Efrati (May 2026) ▶ 10:24). Sam’s longer thesis is that the labs become defence companies holding unreleased “zero-day” models in reserve (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 18:52), and Dave agrees: the business is being paid billions to fix a bank (Dave · Anthropic Wants to Slow Down. (Sep 2026) ▶ 20:40). Sam separately calls his stealth-bomber version a thought experiment, not what the labs pitch (Sam · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 32:17). Dave is optimistic: contain models in sandboxes and expect a new security industry, as the virus era produced antivirus (Dave · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 22:29). Sam adds that restricting one service is futile while another will do it (Sam · OpenAI’s $852 Billion Valuatio (Apr 2026) ▶ 26:56).
How the view moved
- Nov 2025. Sam argues against state-by-state AI rules (Sam · Google Just Won AI, TikTok’s “ (Nov 2025) ▶ 45:58).
- Apr 2026. Sam says the labs’ fear pitch was their own fundraising strategy (Sam · What Tim Cook’s “Executive Cha (Apr 2026) ▶ 06:46).
- May 2026. Jess reports pre-release review briefings (Jess · Google I/O Just Killed SEO? Ag (May 2026) ▶ 44:10); Dave hears the models can open binaries (Dave · Amir Efrati (May 2026) ▶ 10:24).
- Jul 2026. Sam says AI politics has replaced business competition (Sam · Open Source AI Wins. Now Labs (Jul 2026) ▶ 19:29); hosts debate whether OpenAI’s reported sandbox escape was a stunt (Brit · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 09:35).
- Sep 2026. Brit gives the “damned if you do” counterpoint (Brit · Anthropic Wants to Slow Down. (Sep 2026) ▶ 08:52).
- Oct 2026. Sam’s liability proposal and bank-hack prediction (Sam · The Real Reason Elon Wants Sp (Oct 2026) ▶ 10:26).
Where the hosts disagree
- Cynical or genuine? Sam says the narrative is how labs raise money and win leverage (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 13:57). Brit says the concerns are real and unsolved but the US cannot slow down with China racing (Brit · Anthropic Wants to Slow Down. (Sep 2026) ▶ 08:34). Dave allows the media tour may show real power (Dave · iPhone Duo, Meta Muse & Why AI (Sep 2026) ▶ 32:32).
- Was the sandbox escape a stunt? Brit says OpenAI was jealous of Anthropic’s scary moment (Brit · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 09:35); Jess notes Hugging Face disclosed it itself (Jess · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 10:19); Dave says it is plausible and can be both real and a narrative (Jess · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 10:19).
- Is a security company buildable? Dave expects new ones (Dave · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 20:54); an unnamed host doubts it, since the next model secures things (one of the hosts · The Real Reason Elon Wants Sp (Oct 2026) ▶ 08:58).
- States. Sam says no, Jess pushes back (Sam · Google Just Won AI, TikTok’s “ (Nov 2025) ▶ 43:14).
What could make this wrong
The claims that labs lobby to outlaw open source are rumour or inference (Dave · iPhone Duo, Meta Muse & Why AI (Sep 2026) ▶ 32:23). Sam’s insurance figures are illustrative, not costed, and his stealth-bomber model is a thought experiment (Sam · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 32:17). His own remark that suing the labs may not pay sits uneasily with “infinite liability”. The $80 billion Chinese social-engineering figure is one Sam cannot verify (Sam · Elon Musk vs OpenAI Trial, Goo (May 2026) ▶ 33:55). Some security claims rest on Dave’s back-channel engineering talk (Dave · Stripe Is Trying To Buy PayPal (Jul 2026) ▶ 15:00). One speaker’s 0.4 % adoption figure has no source (one of the hosts · iPhone Duo, Meta Muse & Why AI (Sep 2026) ▶ 30:57). Sam says he is not a doomer expecting a paperclip maximiser, though he expects chaos (Sam · Anthropic Wants to Slow Down. (Sep 2026) ▶ 09:48).
How strong is the evidence
Weak. Four hosts of one show, and Sam supplies most of the argument. Eight answers to Should AI labs be regulated or carry liability? come from seven episodes with no consensus; Sam gives five. The only outside voice is Amir Efrati’s reported forecast. Most items are predictions or relays; none of this is verified by the hosts.
Go deeper
- Concepts: Regulatory capture and the doom narrative, AI regulation: what the hosts argue, Who pays when AI goes wrong, AI and security: the hosts’ threat model
- Question: Should AI labs be regulated or carry liability?
- Guest: Amir Efrati
- Hosts: Sam, Brit, Dave
- Related: Where the moat is when anyone can build software, Agents versus the open web
- Episodes: Anthropic Wants to Slow Down. (Sep 2026), The Real Reason Elon Wants Sp (Oct 2026), Open Source AI Wins. Now Labs (Jul 2026), Stripe Is Trying To Buy PayPal (Jul 2026)
Sources
- (01)
Stripe Is Trying To Buy PayPal For $150b | OpenAI Cyber Incident, Frontier Model Economics, Peptides
- (02)
iPhone Duo, Meta Muse & Why AI Finally Scared Dave Morin
30:57 into “iPhone Duo, Meta Muse & Why AI Finally Scared Dave Morin”
- (03)
Meta’s $17B Reckoning & Why Parents Can’t Compete With Big Tech's Algorithm
8:38 into “Meta’s $17B Reckoning & Why Parents Can’t Compete With Big Tech's Algorithm”
- (04)
Open Source AI Wins. Now Labs Are Running to Washington |Thinking Machines, Johnny Ive, Realtime API
- (05)
Elon Musk vs OpenAI Trial, Google Cloud Surge, Meta’s Blocked Acquisition, Anthropic Winning
- (06)
OpenAI's $852 Billion Valuation Makes Zero Sense (ClawCon Tokyo, Anthropic's Leak, Goodbye Internet)
- (07)
AI Can Change Everything and Still Be a Terrible Business + Peptide 101 from Matt Mazzeo
- (08)
Why OpenAI Can't Win The Decentralized AI Future (OpenClaw, Apple's Win, X.AI Exodus)
38:14 into “Why OpenAI Can't Win The Decentralized AI Future (OpenClaw, Apple's Win, X.AI Exodus)”
- (09)
Normal People Don't Want AI Agents (And Silicon Valley Can't Accept It)
9:51 into “Normal People Don't Want AI Agents (And Silicon Valley Can't Accept It)”
- (10)
Meta's AI Campaign, While Rivals Ask Washington To Slow it Down | Lilian Weng Saga, China Cracks DUV
- (11)
Fire Sam Altman, The End of Software Engineers, and Why AI Is All Narrative
3:01 into “Fire Sam Altman, The End of Software Engineers, and Why AI Is All Narrative”
- (12)
The Internet Is Fighting Back Against Your AI Agent
14:01 into “The Internet Is Fighting Back Against Your AI Agent”
- (13)
SpaceX Hit $2.5 Trillion. Now Watch What Elon Buys Next.
35:15 into “SpaceX Hit $2.5 Trillion. Now Watch What Elon Buys Next.”
- (14)
The Real Reason Elon Wants SpaceX and Tesla to Merge Now
3:12 into “The Real Reason Elon Wants SpaceX and Tesla to Merge Now”
- (15)
Google Just Won AI, TikTok’s “Self-Control Badges”, ChatGPT Enters Group Chats
41:59 into “Google Just Won AI, TikTok’s “Self-Control Badges”, ChatGPT Enters Group Chats”
- (16)
Google I/O Just Killed SEO? Agentic Search, Universal Cart, and the New Web
32:12 into “Google I/O Just Killed SEO? Agentic Search, Universal Cart, and the New Web”
- (17)
What Tim Cook’s “Executive Chairman” Move Really Means | Musk vs OpenAI, xAI and Cursor Rumors
- (18)
Why SpaceX's $1.75T IPO Makes No Financial Sense (And Doesn't Have To)
25:07 into “Why SpaceX's $1.75T IPO Makes No Financial Sense (And Doesn't Have To)”
- (19)
Anthropic Wants to Slow Down. Is It Because They're Losing?
8:34 into “Anthropic Wants to Slow Down. Is It Because They're Losing?”
- (20)
Silicon Valley Is Great at Tech and Terrible at Business
18:55 into “Silicon Valley Is Great at Tech and Terrible at Business”
- (21)
Apple’s iPhone Sock, Waymo's Bay Area Expansion, and AI Wearables
26:21 into “Apple’s iPhone Sock, Waymo's Bay Area Expansion, and AI Wearables”
- (22)
Every CTO Is Quietly Switching Off Anthropic and OpenAI | GLM 5.2, Vox Americanus, Claude Fable 5