AI Pioneer Yoshua Bengio Raises Alarm Over Restricted Claude Mythos Release
Canadian computer scientist Yoshua Bengio, widely recognized as one of the "Godfathers of AI" for his foundational contributions to deep learning, has voiced significant concerns regarding the limited release of Anthropic's latest artificial intelligence model, Claude Mythos. Bengio argues that the core problem is the concentration of decision-making authority within a single private corporation, which he believes poses a substantial risk to global cybersecurity infrastructure.
Concentration of Power in Private Hands
In a recent interview with Fortune, Bengio emphasized that restricting access to such a powerful system allows one organization to unilaterally determine which companies and countries can fortify their digital defenses against emerging cyber threats. "It doesn't make sense that private individuals are deciding the fate of infrastructure for everyone else," Bengio stated. "What about all the companies and all the countries that didn't get access?"
His remarks come as Anthropic selectively shares Mythos—an AI model capable of identifying thousands of previously unknown "zero-day" vulnerabilities—with a small, primarily US-based group of companies and government entities. Bengio warned that this approach risks excluding large segments of the global ecosystem from critical cybersecurity protections, leaving them vulnerable to sophisticated attacks.
Anthropic's Justification and the Dual-Use Dilemma
Anthropic has defended the controlled rollout by citing the dual-use nature of Mythos. While the model can help identify vulnerabilities and strengthen systems, it could also be weaponized to launch cyberattacks that disrupt critical infrastructure. To mitigate this risk, the company has opted for an initial release to select American technology companies whose platforms underpin widely used systems, while briefing the US government as it prepares to extend access to federal agencies.
However, this move has ignited a broader debate about governance, fairness, and international equity. Several governments and institutions have reportedly sought access to evaluate vulnerabilities within their own systems. For instance, the Bank of England publicly stated that Anthropic had assured UK banks of near-term access. Meanwhile, discussions at the IMF and World Bank spring meetings were dominated by concerns about the model's ability to expose weaknesses in global financial systems, especially given that many regulators and companies outside the US have yet to assess its findings.
Calls for International Oversight and AI Sovereignty
The views of Bengio and other experts suggest that the current scenario necessitates greater international involvement in AI regulation. Bengio has proposed establishing an international authority to oversee the production and deployment of highly sophisticated AI technology. He advocates for governments to impose strict rules and regulations on businesses to prevent the misuse of advanced AI from affecting other nations' infrastructure.
"There needs to be an agency really in charge of overseeing these kinds of decisions," Bengio asserted. "As the power of AI continues to grow, this question of international commitment becomes pressing. There's no reason that it's going to limit itself to attacking US infrastructure or US citizens. So this has to be an international affair."
This debate is fueling a broader push for "AI sovereignty," as countries aim to reduce their dependence on foreign technology providers. Geopolitical tensions and fears that access to critical technologies could be influenced by national interests or policy shifts have amplified these concerns.
US Government Involvement and Broader Implications
The US government is actively seeking to integrate Mythos into its operations. Bloomberg obtained a memo from the White House Office of Management and Budget indicating that several federal departments, including the Department of Defense, the Treasury, and Homeland Security, will begin using a version of the model. This development occurs despite an ongoing legal dispute between Anthropic and the Pentagon over earlier supply-chain risk designations.
Beyond proprietary platforms like Mythos, Bengio also cautioned about the risks associated with open-source AI models. While such technology is often viewed favorably for its transparency and collaborative security improvements, he noted that AI has advanced to the point where it can efficiently search open-source software for vulnerabilities, potentially exacerbating cyber threats.
Including China in Global AI Governance
Bengio emphasized the importance of incorporating China into any global AI governance framework, given the intense race between the US and China to develop advanced AI systems. Although he estimates that Chinese models may lag behind their US counterparts by a few months, Bengio stressed that this gap does not significantly diminish the associated risks, underscoring the need for inclusive international cooperation.
Bengio's criticism highlights a fundamental issue: as AI systems become more powerful, decisions about their deployment and accessibility have worldwide repercussions. He contends that entrusting these decisions to a single company could leave crucial parts of the globe unprotected and concentrate excessive control over vital infrastructure in the hands of a few individuals, calling for urgent, coordinated action to ensure equitable and secure AI development.
