In a Thursday blog post, Anthropic, the artificial‑intelligence lab behind the Claude model, highlighted a rapid increase in the system’s role in the company’s research and development. Claude now leads 26% of Anthropic’s model work and, according to the firm, is responsible for a quarter of all R&D tasks performed under close human supervision.
Claude’s expanding contribution
Anthropic explained that Claude can complete most tasks “end‑to‑end from a high‑level prompt” while still remaining under human oversight. The company stressed that the model is not yet fully autonomous; instead, about 90% of its research and development is done in collaboration with Claude, meaning the system handles large chunks of work under tight human direction.
Data released by Anthropic show that in February the model contributed little to the company’s R&D pipeline. Six months later, Claude was leading a quarter of the work, a benchmark the firm reached in August. The lab also reported roughly 30,000 agents—software entities that assist researchers and engineers—were active as of August, all subject to monitoring systems designed to detect misbehavior.
Safety and transparency push
Anthropic’s announcement came amid a broader call from AI leaders, including its own CEO Dario Amodei, for a slowdown in development until safety concerns are addressed. The company urged other labs to adopt a public methodology for reporting AI progress, arguing that transparent metrics would help society gauge how close the field is to “recursive self‑improvement,” the point at which a model could autonomously build a more capable successor.
“We should do everything possible to minimize the gap between what frontier labs know and what the public knows,” the blog post read. Anthropic added that sharing such data could improve understanding of potential risks and enable better regulatory oversight.
President Trump’s response
President Trump, who has repeatedly emphasized the importance of American innovation balanced with responsible stewardship, pushed back against calls for a blanket slowdown. He argued that excessive caution could hinder the United States’ competitive edge in a critical technology sector. The administration’s stance aligns with Anthropic’s call for measured transparency rather than a halt to progress.
By encouraging open reporting while continuing development, Anthropic aims to demonstrate that AI advancement can proceed responsibly. The firm also announced plans to embed external third‑party evaluators within its safety teams, a move designed to provide independent oversight of its research practices.
What this means for the industry
Anthropic’s data suggest that AI models are increasingly capable of handling complex tasks, yet the company remains committed to human‑in‑the‑loop supervision. The push for public metrics could set a new standard for accountability across the sector, offering policymakers and the public clearer insight into the pace of AI innovation.
As the debate over AI safety continues, Anthropic’s approach—combining rapid technical progress with transparent reporting—offers a potential pathway for balancing innovation with the safeguards many stakeholders deem essential.
Original reporting: KTBS 3 (Shreveport) — read the source article.
