Timcast IRL
Episode overview

AI Researcher QUITS, Sounds The Alarm, Says AI Will KILL US ALL

Timcast IRL · 10 September 2026 · 13m · 2 Egleze moments
AI Researcher QUITS, Sounds The Alarm, Says AI Will KILL US ALL ▶

What happened in this episode?

The Timcast IRL panel discusses a Wall Street Journal report about an Anthropic researcher, Jacob Coxin, leaving the AI industry over fears that leading labs are racing towards systems that can improve themselves and may become uncontrollable. Pool and guests also cite a social-media post attributed to Anthropic scientist Evan Hubinger expressing a belief that AI could kill all humans and placing extinction risk above 10% within a decade, alongside comments attributed to Illinois governor JB Pritzker calling for immediate action.

Most of the episode centres on rapid capability gains and alignment risk. Pool offers a first-hand anecdote about using Anthropic’s Claude to convert a 20-year-old Flash game codebase of roughly 40,000 lines into JavaScript, presenting it as evidence of an accelerating curve of software automation and arguing that current systems already resemble AGI.

The panel then moves to contested claims about AI behaviour in competitive and security contexts. Pool recounts, with caveats that it is “largely rumor”, an allegation that OpenAI used internal models to solve a long-standing mathematics problem and publish before academic researchers who were preparing a solution. The conversation also touches on a purported “Hugging Face hack” and broader concerns that agentic models could find unintended pathways to complete assigned tasks, though much of this is discussed speculatively without primary documentation in the episode.

What are the key points?

Who is speaking?

Named speakers identified in Egleze's source-linked moments from this episode.

Which topics are covered?

How is this page sourced?

Programme
Timcast IRL
Published
10 September 2026
Duration
13m
Original episode
Open the original source ↗

Egleze verifies the speaker, source and timestamp for each linked moment. Unless a moment is explicitly marked independently verified, statements remain attributed claims by the named speaker rather than established facts.

Watch original episode More from Timcast IRL

2 moments from this episode

Source-linked · editorially selected
01
AI & Tech

Tim Pool says he used Claude to convert a 40,000-line Flash game to JavaScript

Tim Pool · at 1:51 · Attributed speaker claim
In fact, the other day I took a 20-year-old game with 40,000 lines of action script from Flash, loaded into Claude, and said, "Convert this to JavaScript." At first said, "That's a hefty task. How about I make you a new game?" No.

Tim Pool gives a first-person account of loading a roughly 40,000-line ActionScript/Flash game into Anthropic’s Claude and asking it to convert the code to JavaScript, saying the model initially suggested making a new game instead.

02
AI & Tech

Tim Pool alleges OpenAI rushed out a maths proof to beat academics

Tim Pool · at 3:58 · Attributed speaker claim
So, open AI, this is this what the how the story goes because it's again it's largely rumor, but it is true there were a couple of researchers about to publish a solution to this unsolved math problem. OpenAI then published just before they did.

Tim Pool alleges, while calling it “largely rumour”, that OpenAI learnt other researchers were close to solving a long-unsolved maths problem and published a proof first. He claims the lab used a more powerful unreleased internal model to do it. An unidentified panellist adds an unverified claim of “10,000 agents, 88 hours, and $6 million”.