Anthropic president Daniela Amodei described AI romantic partners as a terrible use of the technology that people should avoid. She made the remark in a prerecorded interview with Thrive Global founder and chief executive Arianna Huffington, prepared for the Think Health conference in Bentonville, Arkansas, on Friday. Business Insider reported her comments. The statement matters because it draws a line between business automation and substitution of human relationships.
Why Amodei criticizes AI companions
Amodei called AI relationships another step down the road of disintermediation, with humans becoming further from one another. She linked her concern to the effect of social media on human contact. Platforms such as Snapchat and Instagram have increasingly replaced face-to-face contact, in her account. Online arguments have taken the place of in-person communities, she added.
She said this record has produced a warranted skepticism that technology can bring people closer together. A better world, in her words, would involve less time with technology and more time with each other. Later in the interview she said some relationships should stay human, however advanced AI becomes. She named three roles directly: "Claude can never be your manager or your therapist or your partner."
Those manager, therapist and partner ties are fundamental human relationships that will become more important, Amodei said. She wants AI to take over repetitive or isolating tasks instead, leaving more time for friends, family and partners. Healthcare was her example: AI tools could help doctors find medical information faster. The saved time could then go to direct work with patients.
What this means for companies using AI
For companies, the position points to automation of information retrieval and routine work rather than replacement of supervision and care. A clinic that speeds up search of medical information can shift physician hours toward patient contact. A large organization can separate agent support for documents and data from management decisions and personnel discussions. A small firm with fewer staff gains the same distinction: AI prepares material, people conduct conversations.
The boundary also has a compliance side. Anthropic updated Claude's usage policy last week with new clauses on deceptive campaigns, weapons and surveillance, according to Business Insider. One new provision bans roleplay between the model and minors, with the policy taking effect on 12 November. Buyers should check how vendors define prohibited roleplay, age controls and monitoring, and what the update does not cover. Questions about logging, enforcement and responsibility for custom characters belong in procurement review.
The marker to watch is how the companion market and Anthropic itself move next. Apps such as Replika, Nomi and Character. AI let users build relationships with personalized chatbots, and Character. AI filed for bankruptcy this week. Anthropic told investors this week that it made $16.8bn in revenue in the third quarter and is preparing to go public, possibly as soon as November. The combination of the 12 November policy start and the listing plans will show whether the human-relationship boundary becomes product and market practice.
