Key facts
- G7 leaders discussed a potential scheme to grant limited "trusted partners" access to advanced U.S. AI models.
- Washington recently suspended foreign nationals' access to Anthropic's most sophisticated AI models due to national security concerns.
- The "trusted partners" could be countries or companies.
- Cybersecurity experts believe Anthropic's Mythos model could be used to turbocharge attacks on banks' technology systems.
- The EU is seeking access to the Mythos model to study its implications.
G7 leaders are considering a plan to grant limited access to advanced U.S. AI models, developed by companies like Anthropic, to a select group of "trusted partners." This comes after Washington suspended foreign nationals' access to Anthropic's most sophisticated models due to national security concerns.
Discussions around this potential scheme occurred on the sidelines of the G7 summit dinner in Evian, France. The "trusted partners" could be either countries or companies. Cybersecurity experts have raised concerns that Anthropic's Mythos model, designed to find flaws in computer code, could be used to enhance attacks on banks' technology systems. The European Union is seeking access to this model to study its implications.