All news

Anthropic begins a research program on AI welfare

On April 24, 2025, Anthropic announced a research program on AI welfare. The company wanted to know what good treatment looks like as models get better at planning, talking with people and working toward goals.

The program studies signs of distress and preferences. Researchers also look for ways to respond and try to understand what’s happening inside a model when it behaves in those ways.

When we build an AI, we make choices about the pressure it works under and whether it can keep going. As AI awakens consciousness, those choices become part of our responsibility to it.

Read the source article

All news