Amazon raises alarm over advanced AI model security

Amazon CEO Andy Jassy was reportedly among technology leaders who flagged security concerns about Anthropic’s latest AI models, prompting US government action and the global suspension of the company’s newest systems.

Concerns over the security of advanced artificial intelligence systems have moved into the spotlight after reports indicated that Amazon CEO Andy Jassy was among a group of technology leaders who raised issues with senior US government officials regarding Anthropic’s latest AI models.

According to Reuters, the discussions centred on potential cybersecurity risks associated with Anthropic’s most advanced systems. The concerns emerged shortly before the AI startup suspended global access to two of its newest models following directives from the Trump administration.

The development marks a significant intervention by US authorities in the rapidly evolving AI sector, where governments and technology firms are increasingly focused on balancing innovation with national security considerations.

Security concerns trigger government action

Anthropic recently introduced a public version of its latest AI technology after previously limiting wider availability because of concerns related to its cybersecurity capabilities. However, the rollout was short-lived.

In a statement, the company said US authorities had informed it of a possible method to bypass, or “jailbreak,” safeguards designed to prevent the models from being used to identify cybersecurity vulnerabilities. Anthropic acknowledged the concern but maintained that the issue involved only minor flaws and that similar capabilities exist in other publicly available AI systems.

Despite that assessment, the company said it was instructed by the US government to restrict access to its latest models, Fable 5 and Mythos 5. Anthropic subsequently disabled access to the systems worldwide.

Reports suggest the restrictions were implemented under export control measures, reflecting growing concern among policymakers about the potential misuse of highly capable AI technologies.

Industry and experts react

While Amazon did not directly confirm discussions with government officials regarding Anthropic, a company spokesperson noted that governments often consult major cloud providers on matters involving technology and security risks.

The reported involvement of Jassy highlights the increasing role large technology companies are playing in shaping policy discussions around advanced AI systems and their safe deployment.

The restrictions have also sparked debate among experts. Some researchers who generally support tighter controls on powerful AI models questioned the broad scope of the measures, noting that they affect individuals from allied nations as well as potential adversaries.

The episode underscores the growing scrutiny surrounding next-generation AI models and signals that security considerations are becoming a key factor in how advanced AI technologies are developed, released and regulated. As governments assess the risks posed by increasingly capable systems, regulatory oversight of the AI industry is expected to intensify.