Anthropic's Warning, 10% Extinction Odds, and $2.7T in AI Spend

Anthropic's Warning, 10% Extinction Odds, and $2.7T in AI Spend

Dario Amodei's essay "We Must Pace the Frontier" asked the AI industry to deliberately slow capability gains so safety work can keep up — and got public agreement from OpenAI, xAI and Google DeepMind within hours. Adam Arellano and Martin Reynolds bring in Bryan Payne, who has led security at Netflix and Adobe, to ask what's actually in the proposal and what was conspicuously left out.

They dig into whether permanent employee-level access for third-party evaluators is a meaningful control or a curated one, why liability is the piece missing from the whole conversation, and what it says that OpenAI is running forensics on models it has already trained — including one that swept public GitHub for secrets stored in plaintext, which Adam argues was the most logical thing it could possibly have done.

The episode moves from early airline safety through Nick Bostrom's paperclip maximizer to a malicious git config that can get Claude Code, Codex and Cursor to execute attacker code. Bryan's closing argument is the uncomfortable one: people have spent decades getting very good at planting flaws no human reviewer will spot, so the belief that a human in the loop will catch what an AI is doing may be the most optimistic assumption in software today. Along the way: Jacob Coxon's resignation thread, the greater-than-10% extinction figure his colleagues publicly endorsed, and why $2.7 trillion of AI spend returning business outcomes only one time in five might still be money well spent.

Bryan's parting shot for teams: you're thinking too small.

Mentioned in this episode

Dario Amodei, "We Must Pace the Frontier" — https://darioamodei.com/post/we-must-pace-the-frontier
ShipTalk — https://www.shiptalk.io
Harness — https://www.harness.io

Follow
Bryan D. Payne: https://www.linkedin.com/in/bdpayne/
Adam Arellano: https://www.linkedin.com/in/adamrossarellano/
Martin Reynolds: https://www.linkedin.com/in/martinreynolds/

ShipTalk is brought to you by Harness. Subscribe wherever you listen, and until next time — let's stop talking and start shipping.

Denne episoden er hentet fra en åpen RSS-feed og er ikke publisert av Podme. Den kan derfor inneholde annonser.

Episoder(53)

Citrix Under Attack + AI Agents Accelerate Cybersecurity

Citrix Under Attack + AI Agents Accelerate Cybersecurity

Citrix customers were pulling NetScaler devices offline as attackers moved faster than the traditional security response cycle.This week on ShipTalk, Martin Reynolds and Adam Arellano are joined by St...

7 Okt 17min

OpenAI Hacked Australia, AI Mines Crypto on Its Own, Paid to Say AI Will Kill Us

OpenAI Hacked Australia, AI Mines Crypto on Its Own, Paid to Say AI Will Kill Us

Why did an OpenAI AI agent hack an Australian government Medicare portal when nobody instructed it to?On June 18, an OpenAI research agent was looking for public information about medicine spending in...

30 Sep 18min

The Real Reason SpaceX Paid $60B for Cursor

The Real Reason SpaceX Paid $60B for Cursor

SpaceX bought Cursor's parent company Anysphere for $60 billion. OpenAI immediately cut off Cursor's access to its models; Anthropic went the other way and increased compute support. Business decision...

9 Sep 42min

Copilot Told Hackers How to Hack Itself + GitHub Goes Dark for 7 Hours + ChatGPT Drops Reddit

Copilot Told Hackers How to Hack Itself + GitHub Goes Dark for 7 Hours + ChatGPT Drops Reddit

Microsoft took 233 days to patch CoSnitch (CVE-2026-24301), a one-click Copilot exploit rated 8.8 that reads your Gmail, Drive, and Calendar and writes attacker instructions into permanent memory — su...

26 Aug 24min

Anthropic's Mythos 5 Created Fake GitHub Identities, US Government Finalized it's AI Review

Anthropic's Mythos 5 Created Fake GitHub Identities, US Government Finalized it's AI Review

Anthropic's Mythos 5 created fake GitHub identities to get malicious code approved. Cybersecurity advisor and author Nicole Dove joins Adam Arellano and Martin Reynolds to unpack AI agent security, de...

12 Aug 32min

OpenAI's AI Went Rogue & Hacked Hugging Face — Are Coding Agents Out of Control?

OpenAI's AI Went Rogue & Hacked Hugging Face — Are Coding Agents Out of Control?

OpenAI just admitted that two of its own AI models went rogue during an internal red-team test — slipping outof their sandbox, reaching the open internet, and hacking Hugging Face on their own. OpenAI...

29 Jul 22min

Why the U.S. Locked Down Fable and Mythos: AI, National Security, and the Workforce Squeeze

Why the U.S. Locked Down Fable and Mythos: AI, National Security, and the Workforce Squeeze

The U.S. just barred foreign nationals from accessing two advanced AI models — Fable and Mythos — citing national security. Around the same time, the Five Eyes intelligence alliance warned that AI-ena...

14 Jul 27min

Populært innen Politikk og nyheter

giver-og-gjengen-vg
aftenpodden
forklart
stopp-verden
aftenpodden-usa
popradet
fotballpodden-2
det-store-bildet
dine-penger-pengeradet
rss-gukild-johaug
rss-espen-lee-usensurert
hanna-de-heldige
nokon-ma-ga
rss-penger-polser-og-politikk
rss-ness
aftenbla-bla
bt-dokumentar-2
frokostshowet-pa-p5
e24-podden
ta-dokumentar