Canadian computer scientist Yoshua Bengio has condemned the concentration of decision-making power within Anthropic, a private company, over its advanced AI model, Claude Mythos, which is capable of identifying thousands of previously unknown “zero-day” vulnerabilities. This private control allows Anthropic to determine which companies and countries can secure their infrastructure against emerging cyber risks, effectively privatizing a critical global defense mechanism.
Bengio, described as one of the Godfathers of AI, stated in an interview with Fortune that it “doesn’t make sense that private individuals are deciding the fate of infrastructure for everyone else.” Mythos, designed to identify critical system weaknesses, is being shared selectively with a small group of primarily US-based companies and government entities.
This controlled release, according to Bengio, risks excluding large parts of the global ecosystem from essential cybersecurity protections. Anthropic justifies its limited rollout by citing Mythos’s dual-use nature, claiming the model could be misused to launch cyberattacks, thus necessitating a controlled distribution.
The company has granted initial access to select American technology companies whose platforms underpin widely used systems and has briefed the US government. Several governments and institutions have reportedly sought access to the model to assess vulnerabilities within their own systems, highlighting the global demand for this privately held technology.
The Bank of England publicly stated that Anthropic had assured UK banks of near-term access, indicating the selective nature of the company's engagement. Discussions at the IMF and World Bank spring meetings were dominated by concerns about Mythos’s ability to expose weaknesses in global financial systems, particularly as many regulators and companies outside the US have yet to evaluate its findings.
Private Capital's Grip on Global Infrastructure
Bengio's criticism underscores how a single private entity wields immense power over global infrastructure security, dictating access to a technology that could protect or expose critical systems worldwide. The selective distribution of Mythos to "select American technology companies" and the US government prioritizes national capital interests and reinforces technological dependence for other nations.
This situation fuels a broader push for “AI sovereignty,” as countries seek to reduce their dependence on foreign technology providers, driven by geopolitical tensions and fears that access to critical technologies could be influenced by national interests or policy shifts.
The State's Role in Securing Corporate Power
The US government is actively working to integrate Mythos into its state apparatus, with a Bloomberg-obtained memo from the White House Office of Management and Budget confirming that federal departments, including the Department of Defence (War), the Treasury, and Homeland Security, will begin using a version of the model. This integration proceeds despite an ongoing legal dispute between Anthropic and the Pentagon over earlier supply-chain risk designations, demonstrating the state's imperative to secure advanced technology for its own strategic and economic ends, even when in conflict with the private entity.
The state's actions ensure that critical AI capabilities remain within the orbit of national capital, serving to project military and economic power. Bengio also cautioned about the dangers of using open-source AI models, noting that AI has become advanced enough to search open-source software for vulnerabilities, further complicating the landscape of digital security.
Reform Efforts Miss the Root of Private Control
Bengio advocates for greater international involvement in AI regulation, suggesting an international authority to oversee the production and use of sophisticated AI technology. He also proposes that governments impose strict rules and regulations on businesses to prevent misuse of advanced AI from affecting other nations’ infrastructure.
Bengio emphasized the importance of including China in any global AI governance framework, acknowledging the ongoing race between the U.S. and China in developing advanced AI systems, though he estimates Chinese models may lag behind US counterparts by a few months. However, these proposals for regulation and international cooperation do not challenge the fundamental private ownership and control of such critical technologies, which remains the source of concentrated power.
The current system allows private individuals and corporations to make decisions affecting global infrastructure, leaving important parts of the world unprotected and concentrating power over vital systems in the hands of a few.