In June, the US government forced Anthropic to disable two of its advanced AI models, Fable 5 and Mythos 5, for all customers without prior notice. The incident, which lasted less than three weeks, has raised significant concerns about AI provider dependency and the importance of governance and contingency planning for organizations.
TL;DR
- Anthropic's sudden AI model shutdown highlights the risks of dependency on a small number of AI providers.
- 54% of organizations adopted AI too quickly and now face challenges in scaling it back or implementing it more responsibly.
- Strong AI governance frameworks are essential for organizations to maintain visibility and control over their AI usage.
What happened
In June, the US government issued an export control directive that forced Anthropic to disable Fable 5 and Mythos 5 for all customers due to national security concerns. The shutdown lasted less than three weeks, but the incident has raised significant questions about AI provider dependency and the need for better governance.
The AI market is highly concentrated, with a small number of providers offering models that many businesses rely on. This concentration of risk has rarely been stress-tested until the Anthropic incident.
The incident has demonstrated that AI tools considered stable and compliant can be subject to export controls, licensing restrictions, or government intervention quickly and without notice.
Why it matters
Organizations need to have visibility into their AI usage and the processes to act quickly in case of an access outage. Mature AI governance frameworks can provide this visibility and help guide employees through disruptions.
Strong AI governance is often resisted by organizations worried about it hindering innovation. However, evidence shows that organizations with robust governance infrastructure are better positioned to adopt AI quickly and responsibly.
The Anthropic incident serves as a prompt for boards to consider AI risk as a critical business risk. It requires the same quality of analysis as other critical dependencies, with resilience, diversification, and exit planning built in from the start.
Key facts
- Anthropic disabled Fable 5 and Mythos 5 for all customers in June due to a US government export control directive.
- The shutdown lasted less than three weeks, but the incident has raised significant concerns about AI provider dependency.
- 54% of respondents in the ISMS The State of Information Security Report 2025 say they adopted AI technology too quickly and now face challenges in scaling it back or implementing it more responsibly.
- 37% of respondents in the report cited shadow AI as a concern, which can lead to increased risk profiles that are difficult to track and reverse.
- 95% of respondents to the survey say they are investing in AI governance and policy enforcement, representing a major shift in how the industry thinks about the relationship between governance and adoption.
- Frameworks such as ISO 42001 provide a structured approach to AI management and governance, offering organizations a way to move beyond ad hoc responses to a defined, continuously reviewed set of controls.
Context
The AI market is highly concentrated, with a small number of providers offering models that many businesses rely on. This concentration of risk has rarely been stress-tested until the Anthropic incident.
The incident has demonstrated that AI tools considered stable and compliant can be subject to export controls, licensing restrictions, or government intervention quickly and without notice.
Strong AI governance frameworks are essential for organizations to maintain visibility and control over their AI usage. This includes clear policies, risk management processes, human oversight, and contingency planning.
