How OpenAI, Anthropic, and other AI startups are pursuing recursive self-improvement, in a bid to build AI that can improve itself with little to no human input
Industry chiefs say technology within their grasp is the key to superintelligence. Safety experts say we're not ready.
Lead Source
How this story grew
Discussion
TechSnif Coverage
AI Labs Race to Build Systems That Improve Themselves
OpenAI, Anthropic, and AI startups are chasing recursive self-improvement — AI that upgrades itself without human help.
The biggest names in AI are locked in a race toward recursive self-improvement — the idea that AI systems could autonomously make themselves smarter, faster, and more capable with minimal human involvement.
OpenAI, Anthropic, and a crop of AI startups are all actively pursuing the approach. Industry leaders claim the technology is within reach and represents the critical path to superintelligence.
Safety researchers disagree. They warn the field isn't prepared for the risks that come with self-improving AI systems — models that could evolve beyond human understanding or control.
The tension is familiar but escalating. The companies building the most powerful AI on the planet say the breakthrough is close. The people studying what could go wrong say guardrails aren't remotely ready. Both sides agree on one thing: recursive self-improvement changes everything.