Anthropic rolls out 'dreaming' to let Claude learn from chats
Technology
Anthropic has rolled out a new "dreaming" feature for its Claude AI agents, letting them look back at their past chats and actually learn from them.
The company says this helps the AI spot patterns it would otherwise miss, making its memory sharper and more useful.
Anthropic publishes Claude constitution
Right now, "dreaming" is in research preview and can update the agent's memory automatically or with a developer's OK, so things stay accurate.
Anthropic is also doubling down on responsible AI, with efforts like publishing a constitution for Claude and even keeping an older model around as a blogger to keep it active despite being put out to pasture.