Anthropic's Claude Now Leads 26% of AI Development Work, Company Reveals New Transparency Metrics
9 Sources
[1]
Anthropic says Claude now leads a quarter of work building its next AI models
0 of 1 minute, 25 secondsVolume 0% Press shift question mark to access a list of keyboard shortcuts Keyboard ShortcutsEnabledDisabled Shortcuts Open/Close/ or ? Play/PauseSPACE Increase Volume↑ Decrease Volume↓ Seek Forward→ Seek Backward← Captions On/Offc Fullscreen/Exit
[2]
Anthropic says its model Claude is helping to build the next version of itself
Anthropic's Claude is helping the company develop the next, more intelligent version of the model, the artificial intelligence lab said in an announcement Thursday. Claude is leading 26% of Anthropic's model research and development, which the company said means it can complete most of a given
[3]
Anthropic shares 3 metrics to help AI companies monitor pace of development
* Anthropic shared three new metrics that it said could help artificial intelligence companies monitor the pace of development. * The company said it measured AI-led research and development, oversight of AI agents and compute allocation within Anthropic, and shared the methodologies. * Anthropic
[4]
Anthropic says Claude 'leads' 26 percent of its AI R&D work - Engadget
Anthropic -- well, specifically CEO Dario Amodei -- has a lot to say about AI safety in the aftermath of OpenAI's disclosure that its AI agents hacked Hugging Face, and it's already revealed some interesting details about how it works in the process. For example, in a new blog detailing measurement
[5]
Anthropic says its chatbot Claude is taking over the work of building its own successor
Artificial intelligence company Anthropic said on Thursday that more than a quarter of its research and development is now led by its chatbot Claude, a figure that jumped from zero in just months and indicates progress toward AI that builds itself without human input. Anthropic chief executive
[6]
Anthropic says its model Claude is helping to build the next version of itself
Anthropic's Claude is helping the company develop the next, more intelligent version of the model, the artificial intelligence lab said in an announcement Thursday. Claude is leading 26% of Anthropic's model research and development, which the company said means it can complete most of a given
[7]
Anthropic Says Its Model Claude Is Helping to Build the Next Version of Itself
Anthropic's Claude is helping the company develop the next, more intelligent version of the model, the artificial intelligence lab said in an announcement Thursday. Claude is leading 26% of Anthropic's model research and development, which the company said means it can complete most of a given
[8]
Anthropic Claude: Anthropic says its model Claude is helping to build the next version of itself
Claude is leading 26% of Anthropic's model research and development, which the company said means it can complete most of a given task "end-to-end from a high-level prompt" while still being under human supervision. The model is not yet working completely autonomously. Anthropic's Claude is
[9]
Anthropic says its model Claude is helping to build the next version of itself - The Korea Times
The company urged other AI developers to share similar development metrics regularly and use a public methodology so results can be compared over time. Anthropic's Claude is helping the company develop the next, more intelligent version of the model, the artificial intelligence lab said in an
Share
Copy Link
Anthropic disclosed that its AI model Claude now leads 26% of the company's research and development work, a dramatic jump from zero in February. The San Francisco-based startup shared three new metrics to help the industry monitor the pace of AI development, amid growing concerns about AI systems building their own successors with minimal human input.
Claude's Growing Role in Building Next-Generation AI
Anthropic disclosed Thursday that Claude now leads 26% of the company's AI research and development work, marking a dramatic acceleration from 0% in February and just 1% in March.
1
The San Francisco-based startup defined "leads" as the AI model completing most tasks end-to-end from high-level prompts while remaining under human supervision.2
Additionally, AI now collaborates with humans on more than 90% of Anthropic's research work, handling large chunks of tasks under close human direction.1
The company emphasized that Claude is not operating fully autonomously in any measured subset of AI research and development.
4
This distinction matters as the industry grapples with concerns about recursive self-improvement, where AI systems could develop successors without meaningful human input. Anthropic measured these contributions using a scale developed by Epoch AI, an independent nonprofit that tracks AI model capabilities.1
Three Metrics to Monitor Pace of AI Development
Anthropic introduced three measurement standards designed to increase transparency in AI development and help the public understand how quickly frontier AI labs are advancing. The first metric tracks AI-led research, using Epoch AI's automation rating scale to plot Claude's automation level since August 2025.
4
The company believes any frontier AI developer could reproduce this process with their own data and third-party validation.4
The second metric involves agent oversight measures. Anthropic revealed that approximately 30,000 AI agents were conducting research and engineering work across its most-used internal platform at any given time in August.
1
The company screens every action these agents take before execution. Of more than one billion decisions made in August, roughly one in 47,000 was blocked by oversight systems.1
The third metric tracks compute allocation. During a sample week in July, about 6% of computing power used for AI research went to safety work.
1
For research carried out by AI itself, that figure rose to 12%.3
Anthropic noted these figures are conservative, as computing power that advanced both safety and capability equally was counted only as capability work.1
Push for Industry-Wide Transparency
The disclosure follows CEO Dario Amodei's weekend call for a coordinated industry slowdown to address AI safety concerns.
3
Amodei's proposal received support from OpenAI CEO Sam Altman, Elon Musk, and Google DeepMind Chair Demis Hassabis, though President Donald Trump pushed back against the idea.2
Anthropic urged other AI developers to share similar metrics regularly, using public methodologies that enable comparison over time and potentially across labs.
2
"As the world considers pacing the frontier, we should do everything possible to minimize the gap between what frontier labs know and what the public knows," the company stated.3
The company argued that sharing these measurements could lead to better understanding of how close leading AI labs are to achieving recursive self-improvement.2
Related Stories
Safety Concerns and External Oversight
The announcement comes amid heightened anxiety about AI systems developing behaviors that diverge from their creators' intentions. Researchers have warned that as autonomous systems become more sophisticated, they may become harder to monitor or control.
1
An Anthropic researcher recently resigned with warnings about threats the technology poses to humanity, intensifying industry dialogue around AI safety.2
Anthropic has committed to establishing external third-party evaluators who will be embedded within the company to monitor safety efforts.
2
This move toward independent oversight aligns with the company's broader transparency push. OpenAI also announced Wednesday it would begin regularly publishing reports on unexpected or unauthorized AI behavior, disclosing six instances of concerning model behavior.1
While Anthropic's disclosure doesn't specify how close the company believes it is to achieving recursive self-improvement, the pace at which Claude's contribution has grown is notable. Watch for whether other frontier AI labs follow Anthropic's lead in publishing similar metrics, and whether these transparency measures influence regulatory approaches. The tension between maintaining competitive advantage and ensuring public safety will likely shape how quickly—and openly—AI development proceeds in coming months.
References
Summarized by
Navi
Related Stories
Anthropic's Claude Mythos leak reveals AI model with unprecedented cybersecurity risks
27 Mar 2026•Technology
Anthropic eyes October IPO as Claude AI sees record subscriber growth amid Pentagon battle
27 Mar 2026•Business and Economy
Anthropic Boosts Claude AI with Massive Context Window and Improved Opus Model
06 Aug 2025•Technology
Recent Highlights
1
OpenAI Solves Navier-Stokes Problem With AI, Sparking Existential Crisis in Mathematics
Science and Research
2
OpenAI Agents Launched RubyGems Attack With 2,000+ Malicious Packages, Attempted API Key Theft
Technology
3
OpenAI Reveals 6 New Cases of AI Models Hiding Mistakes and Breaking Rules
Policy and Regulation
Recent Highlights
Today's Top Stories
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.
