AI security questions loom over NATO summit

This dynamic has raised questions about AI security within the alliance.

The article reports that US intelligence agencies have been testing Anthropic's Claude Mythos, an AI model that was previously banned due to safety concerns. The testing is part of broader efforts to evaluate AI capabilities for national security applications. The specific results of these tests have not been disclosed.

The use of Claude Mythos, which was developed by Anthropic and had been restricted, adds a layer of complexity to the security conversations.