AI models capable of major cyber attacks on governments, business could arrive within months, Five Eyes warns
Five Eyes agencies have warned that advanced AI models could sharply expand cyber attack capabilities within months. The statement puts cyber risk in the boardroom as scrutiny grows over Anthropic's restricted models.
Weeks after the US government ordered Anthropic to suspend access to its newest and most powerful AI models, Claude Fable 5 and Mythos 5, for all foreign nationals, including Indians, intelligence agencies from the Five Eyes alliance have issued a fresh warning in a rare joint statement. According to the alliance, powerful AI models capable of launching devastating cyber attacks on governments and businesses are not years away. They could arrive within months.
The Five Eyes (FVEY) is an intelligence-sharing alliance made up of Australia, Canada, New Zealand, the United Kingdom and the United States. The group rarely issues public warnings of this nature, making the latest statement particularly notable.
AI may defend systems, but it can also attack them
The statement, issued late on Monday night Sydney time, acknowledged that artificial intelligence will eventually strengthen cyber defences. However, it warned that the same technology is also making cyber threats faster, larger and more sophisticated.
"Frontier AI models are anticipated to exceed current industry expectations, fundamentally transforming both offensive and defensive cyber capabilities. The timeline is not years, it is months," the Five Eyes agencies said.
The alliance warned that businesses can no longer treat cyber security as a problem for IT departments alone.
"In this environment, cyber resilience is integral to advancing business continuity, market confidence, and long-term value," the statement said.
"A whole-of-organisation and whole-of-society response is required."
The agencies added that cyber risk is now a leadership issue and should be treated as a core business risk rather than a purely technical challenge.
Why are intelligence agencies worried?
The Five Eyes statement does not mention any AI company or model by name. However, many observers believe the warning comes at a time when governments are increasingly worried about advanced AI systems capable of discovering weaknesses in computer networks.
Much of the attention has centred on Anthropic's latest AI models — Fable 5 and Mythos. Fable 5 is described as a more community-friendly version of Mythos, which was released earlier this year. Mythos is a powerful AI model capable of detecting vulnerabilities in cyber systems. Because of concerns that its capabilities could be misused, the US government recently suspended access to both Mythos and Fable 5 for "foreign nationals", citing advice from national security authorities. The move affected users from several countries, including India.
Meanwhile, a report by Wired suggested that South Korean telecom giant SK Telecom's access to Mythos may have contributed to Washington's decision. According to the report, days before Anthropic took its advanced AI models offline, the White House asked the company to revoke SK Telecom's access to Claude Mythos. US officials were reportedly concerned about what they described as the telecom firm's alleged ties to China.
The Wired report also claimed that concerns about SK Telecom were compounded by separate warnings raised by Amazon. Researchers at the cloud company reportedly found ways to bypass some safeguards built into Fable 5, the public version of Mythos that was released on June 9.
They argued that these weaknesses could potentially expose some of Mythos' advanced cyber capabilities. Anthropic and external cyber security experts reportedly maintained that such risks were not unique to Claude models.
Weeks after the US government ordered Anthropic to suspend access to its newest and most powerful AI models, Claude Fable 5 and Mythos 5, for all foreign nationals, including Indians, intelligence agencies from the Five Eyes alliance have issued a fresh warning in a rare joint statement. According to the alliance, powerful AI models capable of launching devastating cyber attacks on governments and businesses are not years away. They could arrive within months.
The Five Eyes (FVEY) is an intelligence-sharing alliance made up of Australia, Canada, New Zealand, the United Kingdom and the United States. The group rarely issues public warnings of this nature, making the latest statement particularly notable.
AI may defend systems, but it can also attack them
The statement, issued late on Monday night Sydney time, acknowledged that artificial intelligence will eventually strengthen cyber defences. However, it warned that the same technology is also making cyber threats faster, larger and more sophisticated.
"Frontier AI models are anticipated to exceed current industry expectations, fundamentally transforming both offensive and defensive cyber capabilities. The timeline is not years, it is months," the Five Eyes agencies said.
The alliance warned that businesses can no longer treat cyber security as a problem for IT departments alone.
"In this environment, cyber resilience is integral to advancing business continuity, market confidence, and long-term value," the statement said.
"A whole-of-organisation and whole-of-society response is required."
The agencies added that cyber risk is now a leadership issue and should be treated as a core business risk rather than a purely technical challenge.
Why are intelligence agencies worried?
The Five Eyes statement does not mention any AI company or model by name. However, many observers believe the warning comes at a time when governments are increasingly worried about advanced AI systems capable of discovering weaknesses in computer networks.
Much of the attention has centred on Anthropic's latest AI models — Fable 5 and Mythos. Fable 5 is described as a more community-friendly version of Mythos, which was released earlier this year. Mythos is a powerful AI model capable of detecting vulnerabilities in cyber systems. Because of concerns that its capabilities could be misused, the US government recently suspended access to both Mythos and Fable 5 for "foreign nationals", citing advice from national security authorities. The move affected users from several countries, including India.
Meanwhile, a report by Wired suggested that South Korean telecom giant SK Telecom's access to Mythos may have contributed to Washington's decision. According to the report, days before Anthropic took its advanced AI models offline, the White House asked the company to revoke SK Telecom's access to Claude Mythos. US officials were reportedly concerned about what they described as the telecom firm's alleged ties to China.
The Wired report also claimed that concerns about SK Telecom were compounded by separate warnings raised by Amazon. Researchers at the cloud company reportedly found ways to bypass some safeguards built into Fable 5, the public version of Mythos that was released on June 9.
They argued that these weaknesses could potentially expose some of Mythos' advanced cyber capabilities. Anthropic and external cyber security experts reportedly maintained that such risks were not unique to Claude models.
