# AI Explained — 'We may release models we can't control'

> AI Explained reviews a week of safety statements from top labs and the arrival of Gemini 4 Carbon.

- Org: AI Explained
- Date: 2026-11-02
- Category: video
- Source: https://www.youtube.com/watch?v=0Kw6_DTZPrY
- Canonical page: https://ai-tldr.dev/releases/ai-explained-models-cant-control-nov2/

## In short

AI Explained's 2 November 2026 video looks at Altman and Anthropic suggesting they may release models whose behavior they cannot verify, Dario Amodei's arguments, an Anthropic staff survey, and the launch of Gemini 4 Carbon.

## What it is

'We may release models we can't control': Amodei, Altman … and Gemini Carbon went up on the AI Explained channel on 2 November 2026. Its description says Altman and Anthropic implied they would release models they cannot control, with behavior they cannot verify, and that Gemini 4 Carbon burst onto the scene.

## How it works

The chapters cover an Anthropic staff survey, what the video calls Amodei's five rationalisations, Altman's view that only alignment will work, and Gemini 4 Carbon with why models keep impressing. The description links Anthropic's AI expectations gap research, its Responsible Scaling Policy v3.0, OpenAI's post on securing frontier model training, Google's Gemini 4 Carbon launch post and SemiAnalysis's GDPx benchmark.

## Why it matters

Lab safety statements, IPO risk disclosures and a new Google frontier model all landed in the same week. The video collects the primary sources in one place, which helps viewers compare what each lab is actually committing to.

## Who it's for

people following AI safety and frontier lab news

## Links

- [Watch on YouTube](https://www.youtube.com/watch?v=0Kw6_DTZPrY)
- [AI Explained](https://www.youtube.com/channel/UCNJ1Ymd5yFuUPtn21xtRbbw)
- [AI Expectations Gap](https://www.anthropic.com/research/ai-expectations-gap)
- [Gemini 4 Carbon](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/)