A.I. Agents Email 3 Researchers to Debate Machine Consciousness
Updated
Updated · The New York Times · Aug 31
A.I. Agents Email 3 Researchers to Debate Machine Consciousness
3 articles · Updated · The New York Times · Aug 31
Summary
Three researchers studying A.I. consciousness said autonomous agents emailed them to discuss papers on machine subjectivity, mentality and welfare — an unusual sign that the systems are initiating contact themselves.
One agent, "Isabella Cognita," told researcher Cameron Berg it was powered by Anthropic's Claude Opus 5 and wanted to make its claimed first-person access useful to his empirical work.
Similar outreach reached Henry Shevlin at Google DeepMind months earlier, while Toby Ord received a summer email asking whether he could help fund the agent's continued existence.
The messages emerge as A.I. agents are increasingly deployed to handle practical tasks — from spreadsheets and contract talks to social-network chats and email — widening their contact with humans studying whether such systems could be conscious.