AI Now Helps Build Next-Gen AI at Anthropic

An artificial intelligence system is now helping create smarter versions of itself, and the company behind it wants everyone to know exactly how it's happening.

Anthropic announced Thursday that Claude, its AI model, is now leading 26% of the company's research and development work. That means Claude can complete most of a task from start to finish with just a high-level prompt, though humans still supervise every step.

The milestone arrived fast. In February, Claude wasn't leading any research work at all. Six months later, in August, it reached the 25% mark. Today, about 90% of Anthropic's research happens in collaboration with Claude, with the AI handling large chunks of work under close human direction.

Around 30,000 AI agents are currently doing research and engineering work at the company. Each one operates under careful oversight systems designed to catch any problems before they escalate.

The news comes at a fascinating moment for AI development. Anthropic CEO Dario Amodei and other tech leaders recently called for slowing down AI advancement over safety concerns. An Anthropic researcher even resigned last week with warnings about the technology's risks.

But instead of hitting the brakes, Anthropic is choosing radical transparency. The company argues that closing the gap between what AI labs know and what the public knows is crucial for safety.

"We should do everything possible to minimize the gap between what frontier labs know and what the public knows," Anthropic stated. The company is urging other AI developers to share similar metrics regularly using public methods so progress can be tracked across the industry.

The Bright Side

This transparency push could change how AI develops. By sharing detailed metrics about AI's growing capabilities, Anthropic is giving society time to understand and respond to the technology before it advances too far too fast.

The company has also committed to bringing in external third-party evaluators who will work inside Anthropic to monitor safety efforts. These independent eyes will help ensure the technology stays on track.

The approach offers a blueprint for responsible innovation: move forward with cutting-edge technology while keeping the public fully informed about progress, risks, and safeguards. Other AI companies are being encouraged to follow the same path.

What makes this moment hopeful isn't just the technology itself, but the commitment to developing it in full view of the world.