AI Business虎嗅

When AI Starts Building AI

虎嗅Oct 14, 20261 min read

AI Summary

On September 28, 22 AI leaders, including three Turing Award winners and co-founders of Anthropic, published a paper on recursive self-improvement, arguing that automating AI R&D could trigger an intelligence explosion. The article traces the idea from Turing and I.J. Good through the AI winters to today's rapid progress.

AI Insights

The 22-signatory paper on AI R&D automation and intelligence explosion signals that frontier labs now treat recursive self-improvement as a near-term engineering target rather than science fiction. For the tech industry, this implies compressed development cycles, massive compute concentration, and urgent pressure on governance and safety frameworks before the window for intervention narrows.

Analysis written by AI from the original article. Check it against the source.

Translated text

In 1940 at Bletchley Park, Turing, while breaking German codes, said machines would one day surpass humans. The term 'artificial intelligence' emerged in 1950, and in 1965 I.J. Good predicted that a superintelligent machine designing better machines would trigger an 'intelligence explosion.' After AI winters, deep learning rose in 2012, AlphaGo defeated Lee Sedol in 2016, and the Transformer was introduced the same year. After ChatGPT launched in November 2022, progress accelerated sharply. OpenAI announced that an automated AI research intern would arrive in September 2026, with an automated AI researcher by March 2028. On September 28, 2025, 22 AI experts published a paper arguing that automating AI R&D could compress a year of progress into five weeks and that the window for action may close once an intelligence explosion begins. Hinton and Bengio expressed concern, and Anthropic co-founder Clark publicly admitted he was very afraid.

Related News

AI BusinessIT之家

Refusing the Ruling: Anthropic Pushes Back Against US Government's AI 'Supply Chain Risk' Designation

Anthropic has asked a US federal appeals court to reconsider a ruling that upheld the Trump administration's designation of the AI company as a supply chain risk to national security. The company argues the panel misinterpreted the law and that the designation is retaliation for its public statements on AI safety. The case will help determine whether the federal government or private companies control safety safeguards for AI technology.

AI BusinessIT之家

OpenAI Saudi-Born Engineer Arrested by FBI for Secretly Installing Recording Device in Executive Conference Room

According to the U.S. Department of Justice, the FBI arrested Abdulrahman Alfozan, a Saudi-born engineer at OpenAI, on charges of secretly installing a USB-sized recording device in an executive conference room at the company's San Francisco office. OpenAI fired the employee and referred the case to federal investigators.

AI Business虎嗅

AI Personal Assistants, AI Phones and Embodied Intelligence Have Yet to Find Real Demand

A Chinese investment columnist outlines Zeng Ming's three-stage model of AI: infrastructure, mass application, and AI-native applications. He argues that only AI Q&A and AI programming have truly monetized so far, while AI personal assistants, AI phones and embodied intelligence remain driven by corporate strategic ambitions rather than genuine consumer demand.