Five Eyes warns powerful AI models may raise cyber risk soon
The Five Eyes alliance, made up of the United States, the United Kingdom, Canada, Australia, and New Zealand, warned that powerful AI models could soon give attackers much stronger cyber abilities. The warning followed a US decision to block from using Anthropic’s Fable and Mythos models. Five Eyes said AI can improve cyber defense, but it can also make attacks faster, larger, and more advanced.
The main concern is that may make it easier for bad actors to find and use security weaknesses. Anthropic’s Mythos was described as a powerful model that can detect weaknesses in cyber systems and is limited to vetted . Fable 5 was described as a more broadly friendly version connected to that same high-risk area.
Australia added Anthropic to its national AI plan in March through a non-binding agreement focused on sharing AI progress with the government and promoting safety.
Key points
- Five Eyes warned that advanced AI could sharply increase cyber risk within months.
- The US blocked from using Anthropic’s Fable and Mythos models.
- AI may help defenders, but it may also make attacks faster and more complex.
- Mythos is linked to finding weaknesses in cyber systems and is available only to vetted .
- Solo makers should track access limits and security rules around advanced AI tools.