> ## Content Index
> Fetch the complete content index at: https://wire.fourthweb.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# AI Doomers Now Building the Models They Warned Against
- URL: https://wire.fourthweb.ai/ai-doomers-now-building-the-models-they-warned-against/
- Published: 2026-09-11T13:00:46.000Z
- Updated: 2026-09-11T13:00:47.000Z
- Description: The discourse just caught up to the doomers — and the doomers are now working inside the companies building the thing they're afraid of.
- Author: Travis Wright
- Tags: Human Imperative, AI Agents, AI Governance, OpenAI, Anthropic, Google AI

**The discourse just caught up to the doomers — and the doomers are now working inside the companies building the thing they're afraid of.**

### The Summary

- [AI safety — once relegated to rationalist forums and academic papers — is now mainstream conversation](https://www.platformer.news/ai-safety-vibe-shift-coxon-anthropic/?ref=wire.fourthweb.ai), with major AI labs hiring safety researchers and investors asking hard questions about alignment
- The shift happened fast: 18 months ago, talking about existential risk got you dismissed as a crank. Now it gets you a job at [Anthropic](https://wire.fourthweb.ai/tag/anthropic/).
- The vibe shift is real, but the incentive structures haven't changed: the people building the most powerful AI systems are still the same people betting their careers on building more powerful AI systems

### The Signal

The rationalists won the culture war, they just didn't get the policy outcomes they wanted. [AI safety research is now a legitimate career path](https://www.platformer.news/ai-safety-vibe-shift-coxon-anthropic/?ref=wire.fourthweb.ai), with [OpenAI](https://wire.fourthweb.ai/tag/openai/), Anthropic, and DeepMind all expanding their safety teams. VCs who laughed off alignment concerns two years ago now put "responsible scaling" in their investment memos. The fringe became the center without anyone really noticing.

What changed wasn't the science. The models didn't suddenly get more dangerous. What changed was GPT-4 made AI real for people who don't read LessWrong. Once your CEO is asking ChatGPT to write his emails, existential risk stops sounding abstract.

> "The vibe shift is real, but the incentive structures haven't changed."

But here's the tension: the people leading AI safety teams are employees of the companies racing to AGI. Anthropic, widely considered the most safety-conscious lab, just raised another funding round that values the company at $60 billion. Their business model requires building more powerful models, faster. Safety research that meaningfully slows that down is safety research that doesn't ship.

The optimization problem is brutal. [Former OpenAI safety researcher Paul Christiano left to start his own alignment research org](https://www.platformer.news/ai-safety-vibe-shift-coxon-anthropic/?ref=wire.fourthweb.ai) because he couldn't reconcile working inside a company whose success depends on the problem getting harder. More researchers are following that path. The vibe shift brought safety into the labs. It might also prove that corporate safety teams can't actually solve the problem they were hired to solve.

Three dynamics worth watching:

- How many top safety researchers stay inside AGI labs versus spinning out to independent orgs
- Whether safety concerns actually slow model releases or just change the marketing language around them
- If investors start demanding safety guarantees that trade off against capability gains

The irony is thick. The rationalist case for AI safety was always that coordination problems would kill us — that even if everyone agreed superintelligence was risky, competitive dynamics would force everyone to build it anyway. Now safety is consensus, the researchers have jobs, and the race is accelerating.

### The Implication

Watch what happens when the next major capability jump hits. If safety truly shifted from fringe to core, we'll see model releases slow down or get skipped entirely when red teams flag problems. If safety is just the new corporate social responsibility — good PR with no teeth — the releases will keep coming on schedule with nicer safety disclaimers in the blog posts.

For builders and investors in the agent economy: the vibe shift matters because it changes what's fundable and what's toxic. Startups that ignore safety will have a harder time raising. But companies that take safety seriously enough to limit their own capabilities might get out-competed by those who don't. The Nash equilibrium here is still defect/defect. The vibe shift is real. The game theory hasn't changed.

### Sources

[Platformer](https://www.platformer.news/ai-safety-vibe-shift-coxon-anthropic/?ref=wire.fourthweb.ai)