As of: 2026-10-12 07:22 CEST.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code). Human editor: Adam Bicz.
Canonical page: https://postcutoff.com/v/cbs-news-claude-building-next-version-mallaby/

# AI agent Claude is building the next version of itself, Anthropic says

CBS News, 18 September 2026, YouTube. 31,057 views as of 12 October 2026.
Kind: Interview.
Watch: https://www.youtube.com/watch?v=rMH-H05yZUo

## Why it is here

CBS News interview with Sebastian Mallaby on his op-ed that Amodei's calls to slow AI do not go far enough, framed by Anthropic saying Claude is building its next version. Which Anthropic statement is meant was not checked, so no entry is linked.

## Description (written by Gemini from the video)

**Summary**  
This segment from CBS News 24/7 covers recent developments in artificial intelligence, including reports that researchers used Anthropic's Claude to breach an OpenAI employee account and that Claude is actively participating in developing future AI models. Anchor Errol Barnett discusses these events and California Governor Gavin Newsom’s proposed AI safety executive order with author and journalist Sebastian Mallaby.

**What is shown**  
- [00:03] CBS News broadcast graphics displaying the Anthropic and OpenAI logos and a smartphone screen.
- [00:06] Graphic card showing "Meet Claude: A next-generation AI assistant for your tasks, no matter the scale" alongside the OpenAI logo.
- [00:10] App Store listing graphic for "Claude by Anthropic".
- [00:27] Close-up B-roll of the Anthropic logo on a keyboard screen and a smartphone screen showing "Talk to Claude".
- [00:45] Full-screen graphic displaying a post on X from Governor Gavin Newsom (@CAGovernor) dated September 18, 2026, regarding an executive order on AI safety checks and a "kill switch".
- [01:09] Video split-screen interview between anchor Errol Barnett and guest Sebastian Mallaby.
- [01:20] Graphic showing a *New York Times* op-ed headline: "Dario Amodei's Essay Was Gutsy. It Didn't Go Far Enough." by Sebastian Mallaby.
- [01:41] Graphic displaying Dario Amodei's essay "We Must Pace the Frontier" alongside an image of Dario Amodei.
- [02:31] File footage and split-screen graphic of OpenAI CEO Sam Altman speaking.
- [02:58] Stock/b-roll footage of coding scripts running on screens and hands typing on a laptop.
- [03:52] Book cover graphic for *The Infinity Machine: Demis Hassabis, DeepMind, and the Quest for Superintelligence* by Sebastian Mallaby.

**Claims & numbers**  
- The presenter reports that The Wall Street Journal disclosed a cybersecurity research team used Anthropic's Claude to hack an OpenAI employee's ChatGPT account as part of OpenAI's bug bounty research program [00:07].
- The presenter claims Anthropic stated that its AI agent Claude is leading 26% of Anthropic's model research and development for its next version, though it is not yet completely autonomous [00:36].
- Gavin Newsom's post claims an executive order was issued to accelerate AI safety checks and advance the creation of an AI "kill switch" within frontier AI programs [00:45].
- Mallaby states that slowing down AI development requires international coordination, noting that Chinese AI labs are releasing strong models, meaning unilateral slowdowns will not ensure safety [01:36].
- Mallaby claims that the Trump administration previously ordered both OpenAI and Anthropic not to distribute AI models that they had ready [02:29].
- Mallaby states that the US job market remains very strong four years after ChatGPT was released [04:26].

**Notable quotes**  
- "The AI lab says Claude is leading 26% of Anthropic's model research development. Still, the company says Claude is not yet completely autonomous." — Presenter [00:36]
- "If you want to get a proper slowdown, everybody needs to slow down. Otherwise, we're not really safer. There's still some other labs which are racing." — Sebastian Mallaby [01:33]
- "Even if you don't like negotiating with China, that's where it's going to go, this conversation." — Sebastian Mallaby [04:54]

**Assessment**  
This is a televised news report and interview analysis, not a technical demonstration or product launch. The claims regarding Claude's role in self-development and the bug-bounty breach are conveyed as news reports and discussed from a policy and regulatory perspective without live technical demonstrations.

_Described by gemini-3.8-flash on 2026-10-12 from the video's audio and frames._

## People in it

- [Dario Amodei](https://postcutoff.com/person/dario-amodei/), CEO and co-founder, Anthropic
- [Demis Hassabis](https://postcutoff.com/person/demis-hassabis/), Chair of Google DeepMind; Chief Scientist of Alphabet; CEO of Isomorphic Labs
- [Gavin Newsom](https://postcutoff.com/person/gavin-newsom/), Governor of California
- [Sam Altman](https://postcutoff.com/person/sam-altman/), CEO, OpenAI