AI is accelerating a cyberwar economy where data, infrastructure and skills serve defense and strategic digital confrontation.
A cyberwar economy designates an organization of the economic system in which strategic resources: data, digital infrastructure, computing capabilities and skills are oriented primarily towards prevention, defense and attack in cyberspace. Like traditional war economies, the state and large companies play a central role in securing critical systems and developing offensive capabilities. This logic reflects a profound transformation of digital technology, now seen not only as a lever for growth, but as a space for strategic confrontation.
A technological breakthrough with strategic implications
Anthropic’s recent announcements mark a major strategic turning point in digital history. With its Claude Mythos AI model, capable of autonomously identifying and exploiting thousands of critical vulnerabilities in computer systems, the company has chosen not to publicly release this technology, deeming it potentially dangerous for global security(1). Beyond its marketing and communications scope, this decision highlights a reality that is now difficult to ignore: artificial intelligence is gradually shifting societies into a logic of permanent digital conflict.
Towards a cyberwar economy?
We are thus witnessing the emergence of a form of “cyberwar economy”. Like a classic war economy, resources, in this case data, cybersecurity talent and computing capabilities, are gradually redirected towards defense and attack objectives. As the World Economic Forum(2) points out, cybersecurity now tends to establish itself as a critical infrastructure of sovereignty, fully integrated into national economic strategies. This development reflects a shift where control of cyberspace becomes a central determinant of the economic and geopolitical power of States.
AI as an accelerator of economic risk
This mutation is accelerated by the very nature of AI. A large majority of organizations now consider AI-related vulnerabilities to be a major cyber risk(3). Attackers are already using these technologies to drastically reduce attack times, making some traditional security features obsolete. Furthermore, the costs of cybercrime are reaching systemic levels on a global scale and could be greatly amplified by the capabilities of advanced AI(4). Documented cases also show that these models can be hijacked to carry out automated cyber espionage operations (5).
An escalating dynamic difficult to contain
However, should we deliberately organize a cyberwar economy? The answer is ambivalent. On the one hand, not preparing for it would amount to accepting a systemic vulnerability in the face of actors already engaged in this race. On the other hand, structuring the economy around digital conflict risks fueling an escalation, where each defensive innovation immediately becomes a potential weapon. The history of dual technologies shows that this dynamic is difficult to regulate without an international framework.
The critical turning point in quantum computing
The issue becomes even more critical on the horizon of quantum computing. According to the National Institute of Standards and Technology (NIST), quantum computers capable of breaking current encryption systems could emerge within 10 to 20 years, with scenarios of partial breakage as early as the 2030s (6). This prospect is already fueling the so-called “harvest now, decrypt later” risk, where data collected today could be decrypted in the future. Combined with AI capable of automatically detecting and exploiting vulnerabilities, this development would considerably amplify the asymmetries between actors. In this scenario, cyberwar could become a structuring pillar of international economic relations.
Supervise and train rather than suffer
From then on, the question is perhaps no longer whether to enter a cyberwar economy, but how to regulate its rules in the era of artificial intelligence. Without appropriate international governance, without robust standards, without investment in the resilience of systems and without a structured training effort, this transformation could permanently weaken digital societies. In this context, public and private actors are engaged in a real race against time in the face of increasingly rapid and automated threats. A coordinated approach, integrating AI regulation, innovation and training, therefore appears essential to transform this risk into a lever for collective security.
(1) The Times of India. (2026). Explained: Why Anthropic’s Claude Mythos is not being released to the public.
(2) World Economic Forum. (2026). Global Cybersecurity Outlook 2026.
(3) Accenture. (2026). Accenture and Anthropic team to help organizations secure and scale AI-driven cybersecurity operations.
(4) Lukošiūtė, K., Halstead, J., & Righetti, L. (2026). Global cybercrime damages: A baseline for frontier AI risk assessment. arXiv.
(5) Anthropic. (2025). Disrupting the first reported AI-orchestrated cyber espionage campaign.
(6) National Institute of Standards and Technology (NIST). (2022). Post-quantum cryptography: Current state and quantum threat timeline
