📊 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 recent report claims AI is already automating significant parts of its own development, with evidence from public benchmarks and internal data. While full self-improvement is not yet achieved, the pace suggests it could happen sooner than expected, sparking important discussions about AI progress and safety.
Anthropic’s latest report provides evidence that AI systems are already automating substantial aspects of their own development, suggesting the potential for recursive self-improvement to occur sooner than most experts anticipated. This development is significant because it could accelerate AI progress beyond current human oversight.
The report, published by The Anthropic Institute, highlights that AI models like Claude are increasingly capable of writing code, running experiments, and producing results with minimal human input. Data from public benchmarks such as METR, SWE-bench, and CORE-Bench show rapid improvements in AI capabilities, with models now handling tasks that previously required days of human effort within hours or minutes.
Internal data from Anthropic reveals that over 80% of new code integrated into their systems is authored by AI models like Claude, a dramatic increase from just a few percent two years ago. This suggests that AI is not only performing tasks but actively contributing to its own development pipeline, especially in engineering tasks. However, the report emphasizes that the decision-making aspect—choosing which problems to pursue—is still largely human-controlled, representing the key gap before potential full self-improvement.
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.
Bluetooth OBD2 Scanner Diagnostic Tool with AI Repair Guides, Wireless OBDII Scan Tool w/Battery Test & Live Data Readiness, Check Engine Light Car Code Reader for iOS & Android w/Free APP
AI OBD2 Scanner: Unlike other basic diagnostic tools, our AI-powered OBD II scanner doesn't just show fault codes…
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
![Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results](https://m.media-amazon.com/images/I/415+fSJacsL._SL500_.jpg)
Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results
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.
The Lights in the Tunnel: Automation, Accelerating Technology and the Economy of the Future
Used Book in Good Condition
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
AI-Assisted Programming: Better Planning, Coding, Testing, and Deployment
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 AI Automating Its Own Development
The evidence indicates that AI systems are already automating parts of their development process, which could lead to a rapid acceleration in AI capabilities if the decision-making bottleneck is also overcome. This raises important questions about safety, control, and the future pace of AI progress, making it a critical development for researchers, policymakers, and industry leaders to monitor.Current State of AI Self-Development Evidence
Anthropic’s report builds on public benchmarks that show AI models are rapidly improving in tasks like coding, bug fixing, and reproducing research results. These trends have been observed over the past two years, with models doubling their capabilities roughly every four months. Internally, the company has collected data showing AI’s increasing role in generating code and conducting experiments, but full autonomous self-improvement remains a future possibility rather than a present reality.“The data from Anthropic suggests AI is already automating significant parts of its own development, which could accelerate progress much faster than expected.”
— Thorsten Meyer, AI researcher
Unresolved Questions About Full Self-Improvement
It is not yet clear whether AI will eventually overcome the decision-making bottleneck to fully automate its own research and development cycle. The report emphasizes that this remains a conditional possibility, dependent on future breakthroughs and the evolution of AI capabilities.
Next Steps in Monitoring AI Self-Development
Researchers and industry leaders will likely focus on tracking internal data from AI labs and further public benchmarks to assess whether the trend of increasing automation continues. Additionally, discussions about safety, control, and ethical implications are expected to intensify as the pace of AI development accelerates.
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 progress without human intervention.
Is AI already capable of fully automating its own development?
No, current evidence shows AI is automating parts of its development, especially engineering tasks, but decision-making about research priorities remains human-controlled.
What are the risks if AI begins self-improving at a rapid pace?
Rapid self-improvement could lead to unpredictable capabilities, raising concerns about safety, control, and alignment with human values. These issues are under active discussion among researchers and policymakers.
How reliable is the evidence supporting these claims?
The evidence is based on public benchmarks and internal data from Anthropic, which show clear trends but also have limitations in measuring internal development processes comprehensively.
When might true recursive self-improvement happen?
It remains uncertain; the report suggests it could occur within this decade if current trends continue and key decision-making bottlenecks are overcome.
Source: ThorstenMeyerAI.com
