AI: The pace of progress is outpacing organizations' ability to adapt
On June 4, 2026, Anthropic published a report with the straightforward title, “When AI Builds Itself,” which quickly gained attention beyond expert circles. Beyond the debate it sparks about the future of AI, it poses a much more immediate question to executives: Is your organization keeping pace with the technology it is adopting?
What Really Happened
The report, authored by Marina Favaro and Jack Clark for the Anthropic Institute, puts forward a clear argument: AI is now actively contributing to the development of AI. To support this claim, the report cites previously unpublished internal data: by May 2026, more than 80% of the code integrated at Anthropic is said to have been written by its own model, Claude. The authors introduce the concept of “recursive self-improvement”—that is, an AI capable of designing its own successor with limited human intervention.
Anthropic is careful, however, to set a limit. This threshold has not been reached and is not considered inevitable. The company simply believes that it could occur sooner than institutions are prepared for. Its proposal is measured: to preserve the option of slowing down or temporarily suspending cutting-edge development through a coordinated and verifiable mechanism involving multiple laboratories and countries. The analogy drawn is that of arms control treaties, with an added challenge, since training a model is easier to conceal than an arsenal. This is therefore not a call for a unilateral and immediate halt.
Credibility also requires listening to criticism. Many view this as a form of “fear-mongering” coming from a company valued at around $965 billion, and as a way to exploit the political debate to stifle competition. This interpretation is worth considering. It does not invalidate the underlying message, but it does encourage us to view it with a degree of detachment.
The Real Signal for Businesses
For a leader, the challenge is not to settle this global debate, which is playing out on a scale beyond the reach of any single organization. The key takeaway lies elsewhere: AI is no longer a distant prospect to be observed; it is an operational reality that is accelerating. And this acceleration is creating a gap. The pace at which these tools are evolving now exceeds organizations’ ability to absorb them—that is, to understand them, manage them, and integrate them into their processes and culture.
It is precisely this gap that poses the real risk. Not AI that gets out of control, but organizations that adopt tools faster than they can develop the skills and rules needed to use them responsibly.
The responsibility for developing skills
Management’s responsibility, therefore, is not to slow down technology. It is to ensure that the organization adopts it in a controlled manner. This requires three conditions.
First and foremost, clear governance: who can use which tools, for what purposes, with what data, and under what rules. In the Swiss context, this requirement aligns with expectations regarding data protection and industry compliance.
Next comes a genuine increase in competence. Beyond mere curiosity, a structured process of onboarding teams and managers helps transform experimentation into reliable practice.
Finally, a user-first approach: starting with concrete cases with demonstrable value, rather than deploying technology in search of an application.
This measured and demanding approach is what distinguishes opportunistic adoption from sustainable transformation.
How Darest Supports This Process of Maturity
This is where Darest IT&AI Solutions comes in. Our role is not to promise spectacular AI, but to help organizations adopt it in a methodical way: conducting maturity assessments, defining appropriate governance frameworks, fostering cultural adoption among teams, and supporting deployments that deliver measurable value.
Our belief is simple. The value of AI lies not in the power of the models, but in how well they are integrated into an organization that retains control over them.
Conclusion
Anthropic’s report will continue to be debated, and that’s healthy. But for leadership, the relevant question isn’t whether the world needs to slow down. It’s whether your organization is moving at a pace it truly controls.
If you would like to assess this level of maturity objectively and determine the next steps, we offer an initial consultation or a targeted assessment.