Breaking

Anthropic Says Claude Now Leads 26% of Its AI Research Work

September 18, 2026 at 04:00 EDT

Foundation Models
AI Agents

Anthropic says Claude is now leading 26% of the company's AI research and development work, a share it says was near zero at the start of 2026. The disclosure marks a new data point in how quickly frontier AI labs are using their own systems to build, test and improve the next generation of models.

September 2026 · Anthropic

Claude Takes a Larger Role in AI R&D

Anthropic says Claude now leads 26% of its own AI research and development implementation work, up from nearly zero at the start of 2026.

26%

AI R&D work Claude now leads

80%+

Implementation share required for a task to count

2026

Year the measured rise began near zero

From near zero to 26%

Near 0%

Start of 2026

26%

Current reported share

Each blue segment represents one percentage point of reported AI-led implementation work.

How the work still flows

Human sets goal

Research intent and acceptance stay human-directed.

→

Claude implements

The task counts when Claude leads at least 80% of implementation.

→

Human reviews

People remain accountable for outputs and decisions.

Why it matters

The figure suggests frontier models are becoming useful inside real engineering and research loops.

Key caveat

The metric is self-measured, human-supervised and not yet comparable across labs.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.