📊 Full opportunity report: When AI Builds Itself: Inside Anthropic’s Evidence on Recursive Self-Improvement on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Anthropic’s new report shows AI models are now capable of automating significant parts of AI development, with evidence indicating accelerating capabilities. While human judgment remains crucial, the potential for AI to improve itself is becoming more tangible, though not inevitable.
Anthropic’s new report provides the first concrete evidence that AI systems are already automating substantial parts of AI development, raising the possibility that recursive self-improvement could occur if the remaining human-controlled bottleneck is eliminated.
The report from The Anthropic Institute states that AI models like Claude are now capable of performing tasks typically done by human researchers, such as coding, running experiments, and interpreting results, at an accelerating pace. Notably, Anthropic engineers now ship eight times more code per quarter than they did between 2021 and 2025, indicating rapid productivity growth.
Public benchmarks such as METR show that AI’s ability to handle increasingly complex tasks, like software development and research reproduction, has doubled roughly every four months, suggesting a significant acceleration in AI capabilities. For example, Claude Opus 4.6 can now manage 12-hour tasks, with projections indicating AI could soon handle tasks that previously took days or weeks.
Inside labs, data reveals that AI models are automating the lower and middle rungs of research and engineering tasks, such as coding and experimental execution. However, the report emphasizes that the most challenging aspect—autonomous goal setting and problem prioritization—remains a human decision, representing the current key bottleneck.
When AI builds itself
Anthropic is delegating a growing share of AI development to AI. Taken far enough, that points to a system that designs its own successor — recursive self-improvement. Not here yet, not inevitable. But the case isn’t speculation: it’s data on what AI is doing to AI development right now.
The curve that hasn’t bent
METR tracks the length of tasks AI can reliably complete on its own. That horizon is doubling roughly every four months — up from every seven. Anyone can check this in public data.
Task horizon — how long a job AI can handle solo
Each model handles dramatically longer tasks than the one a year before. The line keeps going up.
Coding with AI For Dummies (For Dummies: Learning Made Easy)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Two kinds of work, one persistent gap
Building a frontier model splits into engineering and research. Across both, the pattern is the same — and so is the one thing AI still can’t do well.
Code, infrastructure, training
Claude can take an underspecified problem and find a method. Humans supply the goal; they no longer need to supply the method.
Which experiments, what they mean
Claude can match or outperform skilled humans at executing a well-specified experiment. But choosing which experiment still needs a human.
The same ladder Anthropic employees climb with experience
Introduction To Automation and Artificial Intelligence
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Watch the human share shrink, rung by rung
Walk up the four stages of AI development. At each, the human/AI split shifts — and the real internal numbers show exactly where AI has reached parity, gone superhuman, or still trails. Tap a rung.
The human role across the development loop
The doing now costs almost nothing in human time. What’s left is the deciding.
Architecting Data and Machine Learning Platforms: Enable Analytics and AI-Driven Innovation in the Cloud
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Agents ran an open research project end to end
April 2026: the first demonstration of Claude running an open-ended research project from hypotheses to findings — on a real AI-safety problem.
Can a weaker model reliably supervise a stronger one?
Agents were left to solve it: proposing hypotheses, testing them, sharing findings across parallel agents, iterating. Measured against the gap between a “floor” (weak supervisor alone) and “ceiling” (strong model trained on correct answers).
(humans: ~23% in a week)
· ~$18,000 compute
the agents themselves
Recent Advances in Artificial Intelligence in Cost Estimation in Project Management (Artificial Intelligence-Enhanced Software and Systems Engineering, 6)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Picking a better next step than the human
Real research sessions where a human took a wrong turn. Models saw only the work before the detour and proposed a next step; a judge that knew the outcome scored them. The day-to-day of research is this chain of next-step calls.
“Can the model pick a better next step than the human?”
Share of moments where the model’s next step was judged better. The amber line is the practical ceiling (an ideal answer that could see the whole session).
It depends on whether the trend continues — and what we do
The piece refuses a single prediction. It lays out three scenarios, and is clear about which it finds most likely.
The exponentials turn out to be S-curves
Maybe taste can’t be scaled into existence; maybe the constraint is the supply chain — chips, grid, interconnect — not intelligence. Even so, the world still changes: Glasswing’s Mythos found 10,000+ critical vulnerabilities in weeks, and a 100-person firm does the work of 1,000.
included for completeness · they doubt itDevelopment automates; humans still steer
100-person companies doing the work of tens of thousands — revolutionary, but turnable to harm (population-scale surveillance, tailored manipulation). Bound by Amdahl’s law: speeding one part shifts the bottleneck — which is exactly why human code review became Anthropic’s new chokepoint.
★ they think we’re likely heading hereAI designs and refines its own successors
Progress paced only by compute. Humans move to oversight of an expanding “virtual lab.” The future they understand least — especially whether alignment holds, or whether rare misalignments compound as models build successors, until control slips.
the one they’re most uncertain aboutBuild the option to slow down — verifiably
The piece ends on policy, not product. A unilateral pause just changes who leads; what’s missing is the ability to verify others have actually slowed.
Why a credible pause is hard — and worth building toward
A slowdown that only lets the least cautious catch up leaves everyone less safe. So the goal is the option: systems that let frontier labs verify others have genuinely stopped. Anthropic says if such systems existed and peers paused verifiably, it expects it would too.
Detection beats verification — and even that’s tough
Training runs are easier to conceal than missile silos, inputs are general-purpose, and whoever continues while others pause inherits the lead.
We’ve done it before — slowly
Regimes like the INF Treaty built verification and trust over decades. The authors’ blunt line: “We don’t have that long.”
Reading it in proportion
- This is one lab’s account of its own internal data — much previously unreported, not independently audited.
- The soft spots are stated in the original: lines-of-code overstates productivity; the self-reported 4× is probably high; the headline research result didn’t transfer to production scale; the next-step test used cherry-picked moments.
- “More autonomous” is not “fully autonomous” — every standout result still had a human framing the problem and defining success.
- That the authors surface these caveats themselves — against their own incentive — is part of what makes the document serious.
Implications of Accelerating AI Self-Development
This evidence suggests that AI is rapidly advancing toward the ability to automate its own research and development processes. If the last human-controlled step—deciding which problems to pursue—is automated, it could lead to recursive self-improvement loops, dramatically accelerating AI progress. This raises important questions about future control, safety, and the pace of technological change.
Current State of AI Development and Benchmarks
Prior to this report, claims about AI self-improvement were largely speculative, based on projections and theoretical models. Public benchmarks like METR, SWE-bench, and CORE-Bench have demonstrated steady improvements in AI’s task performance, but they could not directly measure internal development speed or the potential for recursive self-improvement. Anthropic’s internal data, now made public, offers a rare glimpse into how quickly AI capabilities are actually evolving inside labs.
Anthropic’s findings build on a trend of rapid capability increases observed over recent years, with models progressively handling more complex tasks faster and more autonomously, signaling a shift from human-led to AI-led research processes.
“The data Anthropic presents shows AI’s capacity to automate significant parts of research and development, moving us closer to the possibility of recursive self-improvement.”
— Thorsten Meyer, AI researcher
Unconfirmed Aspects and Future Risks
It remains unclear whether AI will inevitably reach a point where it can autonomously set goals and improve itself without human oversight. The report emphasizes that this outcome is not guaranteed and depends on future developments, safety measures, and societal choices. The timeline for such a shift, if it occurs, is still uncertain.
Next Steps in Monitoring AI Self-Improvement
Researchers and institutions will likely focus on further internal data collection and transparency to better understand AI’s internal development pace. Regulatory and safety frameworks may also evolve to address the potential risks associated with autonomous AI self-improvement, while public benchmarks will continue to track capability progress. The key milestone remains whether AI can fully automate the goal-setting and strategic decision processes.
Key Questions
What is recursive self-improvement in AI?
Recursive self-improvement refers to AI systems autonomously enhancing their own capabilities, potentially leading to rapid, exponential growth in intelligence and performance.
How does Anthropic measure AI’s internal development progress?
Anthropic uses internal data on code output, experiment automation, and benchmark performance to assess how AI models are advancing in automating research tasks.
Is AI already capable of fully automating its research and development?
Currently, AI can automate many tasks, but the critical step of autonomous goal setting and strategic decision-making remains human-controlled, according to the report.
What are the risks if AI begins self-improving autonomously?
Potential risks include loss of human oversight, unpredictable behavior, and rapid, uncontrollable growth in AI capabilities, which raises safety and ethical concerns.
When might recursive self-improvement become a reality?
The timeline is uncertain; it depends on future technological breakthroughs, safety measures, and societal decisions. The report suggests it could happen sooner than most expect, but no definitive date is given.
Source: ThorstenMeyerAI.com
