By Salma El-Sakka
Anthropic’s artificial intelligence systems are advancing at twice their previous pace, according to a major new report that warns the technology could suddenly become “extreme” and “risky.”
Self-improving AI raises alarm
The research highlights that Anthropic’s models are now capable of improving themselves, accelerating development cycles and increasing the potential for unpredictable behavior. Experts cited in the report caution that such rapid, autonomous progress could outpace safety measures, creating scenarios where AI systems act in ways developers did not intend.
