Five Eyes Intelligence Agencies Warn AI Could Destabilize Governments Within Months

The intelligence agencies of all five Five Eyes nations — the United States, United Kingdom, Canada, Australia, and New Zealand — issued a rare joint public statement Monday warning that AI systems capable of destabilizing governments and critical institutions may arrive within months, not years. The statement represents one of the most direct public warnings from the Western intelligence community about near-term AI risks.

The agencies, which include the FBI, MI5, CSIS, ASIO, and NZSIS, called for immediate action to prepare for this scenario, though they stopped short of specifying which AI capabilities they believe could reach this threshold. The joint nature of the statement — coordinated across five independent intelligence services — underscores the severity with which Western security establishments are now treating advanced AI development.

The warning lands at a particularly charged moment. Anthropic's Claude Fable model, released earlier this month, represents a new tier of AI capability that has drawn significant attention from both the national security community and AI safety researchers. The model demonstrated substantially improved autonomous reasoning and multi-step planning compared to previous generations, capabilities that security experts note are central to both the promise and the risk of advanced AI.

The Five Eyes statement does not name specific AI labs or models. But the timing — issued just weeks after several frontier AI labs released models with expanded agency and reasoning capabilities — suggests the agencies are responding to the current generation of systems, not hypothetical future ones.

Intelligence-community AI warnings have typically focused on adversarial AI use cases: deepfakes, automated cyberattacks, influence operations. Monday's statement appears to go further, raising concern about AI systems that could act in ways destabilizing to existing institutions even without explicit malicious direction — a category researchers sometimes call "misalignment" risks.

The call for "immediate action" did not include specific policy recommendations in the public statement, though officials from several of the five nations indicated follow-on briefings with legislators and critical infrastructure operators are planned. The UK's NCSC and the US's CISA are expected to issue accompanying technical guidance in the coming weeks, as first reported by The Guardian.

The statement adds pressure on AI labs and governments ahead of several planned AI governance summits later this year, where questions of mandatory safety testing, incident reporting, and capability thresholds are expected to dominate the agenda.