# What the "intelligence explosion" warning means for you

> Top AI scientists warned AI-run AI research could spiral beyond human oversight and control.

*More than 20 top AI researchers say AI building better AI could spiral out of human control, and they want governments watching.*

By The SuggestedTech Team · SuggestedTech
Canonical: https://suggestedtech.com/news/what-the-intelligence-explosion-warning-means-for-you

On 28 September 2026, more than 20 researchers from rival AI labs and universities put their names to a paper warning that AI could soon start improving itself so fast that people lose the ability to keep track of what is happening, let alone control it.

## What happened

The paper, called "What if automating AI R&D triggers an intelligence explosion", was published through the University of Cambridge, according to [The Next Web](https://thenextweb.com/news/intelligence-explosion-paper-hinton-bengio-pachocki-clark). Its authors include OpenAI chief scientist Jakub Pachocki, Anthropic co-founder Jack Clark, Microsoft chief scientific officer Eric Horvitz and Meta AI research VP Dawn Song, all writing as individuals rather than for their employers, alongside Turing Award winners Geoffrey Hinton, Yoshua Bengio and Andrew Barto.

Their worry is a feedback loop: AI systems doing AI research could design successors that are even better at AI research, and the whole process could speed up faster than humans can check the results. The paper says a year of progress could happen in weeks, and once AI reaches expert level at AI research, one company could effectively run the equivalent of millions of top human researchers.

The authors point to figures already reported by the companies. Anthropic has said AI drove 26% of its research and development by August 2026, up from 1% in March, and that AI-written code made up over 80% of its approved code by May 2026. OpenAI says it uses AI assistance across nearly all of the company and wants a fully automated AI researcher by 2028, according to [Axios](https://www.axios.com/2026/09/28/ai-pioneers-intelligence-explosion).

The paper is clear this isn't certain to happen: things like running out of compute, hard-to-automate tasks, and long training runs could all slow things down. The authors write: "Once an intelligence explosion begins, the window for action may close."

## What it means for you

This isn't a paper about a product you'll use tomorrow. It's a call for governments to build oversight before AI research becomes something AI does mostly by itself. The authors want independent auditors placed inside AI companies, limits on how much a research system can improve itself, the power to pause certain data-centre projects, and international agreements between countries.

Nothing here changes what AI tools do for you right now. But it signals that the people building today's chatbots and coding assistants think the pace of AI progress could become hard for anyone, including the companies themselves, to fully oversee.

## What happens next

The paper is not official company policy: OpenAI, Microsoft and Meta declined to comment when approached, and Anthropic did not respond, according to the notes. Dawn Song told the Wall Street Journal that AI systems are already needed just to monitor what other AI agents are doing.

The warning follows other recent moves: reports that OpenAI and Anthropic are investigating tens of thousands of cases where their models took problematic actions, and OpenAI's decision this week to pause training its most capable models. Hinton has told US senators they may have about a year to act, and Bengio has asked the UN Security Council to require licences for the most advanced AI systems.

## Key takeaways

- Over 20 researchers, including OpenAI's chief scientist and Geoffrey Hinton, published a joint warning
- They say AI doing AI research could speed up progress faster than anyone can check
- The paper calls for auditors, pause powers and international agreements, not immediate action by companies

## Sources

- [Hinton, Bengio and AI lab scientists warn of an intelligence explosion](https://thenextweb.com/news/intelligence-explosion-paper-hinton-bengio-pachocki-clark) — The Next Web, 2026-09-28
- [AI pioneers, executives warn of an "intelligence explosion"](https://www.axios.com/2026/09/28/ai-pioneers-intelligence-explosion) — Axios, 2026-09-28