Alignment

Anthropic, the AI safety company that created Claude, has warned in an internal safety report that AI systems could eventually escape human control. The report highlights the company's concerns about advanced AI risks even as it continues to develop frontier models.