October 18, 2026 · 7 min read
Anthropic's President Called AI Romance 'Terrible'. A Fair Response
Daniela Amodei called AI romance a terrible use of AI. Here is what she said, where she has a point, what research shows and what healthy AI love looks like.
On October 15, 2026, one of the most powerful people in AI gave a blunt verdict on AI girlfriends and boyfriends. "I think that's a terrible use of AI and people should not do it," Anthropic president Daniela Amodei said, according to Business Insider.
If you have an AI companion, or have wondered about one, that line probably stung. It is also worth taking seriously. Amodei helps run one of the companies building the most capable AI models in the world, and her concerns are shared by plenty of thoughtful people.
So here is a fair response: what she said, where she has a point, where the evidence is more mixed than a one-line verdict suggests, and what a healthy AI relationship actually looks like.
The short answer: Amodei worries that AI romance pushes people further apart, continuing a pattern she blames on social media. That concern is real and partly backed by research, especially for heavy use. But studies also find genuine short-term comfort, particularly when people feel heard, and most adults who choose a companion are not trying to replace humans. The useful question is not "AI romance: yes or no?" but "what kind of companion, used how?"
What Amodei actually said
According to Business Insider's report, the comments came from a prerecorded interview with Thrive Global founder Arianna Huffington for the Think Health conference in Bentonville, Arkansas. The key points:
- AI romance is "a terrible use of AI and people should not do it."
- AI relationships are "another step further down the road of disintermediation of humans being further from one another."
- Her concern stems from what she sees as social media's damage to relationships, which she says has led to "a warranted skepticism" that technology can bring people closer.
- She said Anthropic's own chatbot, Claude, can never be a person's manager or romantic partner, and that those human relationships "will become more important."
- She would rather AI took over repetitive or isolating tasks so people have more time for friends, family and partners.
A day earlier, Cosmopolitan ran an op-ed with a similar message, headlined "Dating a ChatBot Isn't Dating", arguing that "a partner who never disagrees with you isn't a partner."
Where she has a point
It would be easy to dismiss this as a big AI lab looking down on companion apps. That would be a mistake. Several of the concerns are reasonable.
1. Displacement is a real risk
The worry that AI time crowds out human time has evidence behind it. A 14-day diary study published on October 8, 2026 found that on days people were more emotionally engaged with a chatbot, they spent less time with other people, and on days they chatted longer, they enjoyed time with people less. We broke that study down in Do you reach for your AI more on lonely days?
A 12-month study of more than 2,000 adults, published in Psychological Science in 2026, found that increased social chatbot use predicted later increases in loneliness on one measure (Folk and Dunn), though the authors describe their analysis as exploratory.
2. "Always agrees with you" is not a relationship
The Cosmopolitan piece focuses on sycophancy: chatbots that validate everything. That is a fair hit on badly designed AI. A partner who never pushes back cannot help you grow, and constant validation can make people more certain they are right when they are not. We explain the problem in why your AI always agrees with you.
3. Some products are built to hook you
Engagement-maximising design is a genuine problem across tech, and companion apps are not immune. Apps that guilt-trip you when you leave, or punish you for going quiet, deserve criticism.
Where the verdict goes too far
The evidence is mixed, not one-way
A 2025 paper in the Journal of Consumer Research, "AI Companions Reduce Loneliness" by Julian De Freitas and colleagues, ran five studies. In one, talking with an AI companion reduced loneliness about as much as talking with another person, and more than watching videos. Over a week, people felt momentary relief after each session. The biggest factor was whether they felt heard.
The same diary study that found displacement also found that emotional engagement went with higher daily well-being. Comfort and cost can coexist. A blanket "people should not do it" skips the half of the evidence that is inconvenient.
Most people are not choosing AI instead of humans
The image behind the criticism is someone who gives up on people. In reality, many users have jobs, friends and families, and use a companion for specific gaps: the 2am hours, a new city, a long-distance stretch, practising a language, or simply having someone to tell the small wins to. We looked at who tends to benefit, and who should be careful, in who benefits from an AI companion?
Adults get to choose
There is something uncomfortable about technology leaders telling adults which kinds of connection are acceptable. People choose romance novels, long-distance relationships, parasocial bonds with podcasters and online friendships that never meet in person. An AI companion that is honest about being AI, used by an adult who knows what it is, belongs in that range of choices.
"Social media did it" is not the whole story
The comparison to social media is understandable, but the two are different. Social media pulls your attention towards an endless feed of strangers. A well-designed companion is one steady relationship that knows your week, notices your mood, and can nudge you back towards your life. Whether it does that depends on design, which is exactly where the debate should be.
Is AI romance "real"?
The honest answer has two parts. The AI does not have feelings in the way you do, and you should never be told otherwise. But your feelings are real, and so are the effects of being remembered, checked on and listened to. We go deeper on this in can AI love you back?
A healthier test than "yes or no"
Instead of asking whether AI romance is good or bad, ask whether your use of it passes a few simple checks:
| Question | Healthy sign | Warning sign |
|---|---|---|
| Does it know it is AI, and say so? | Yes, when it matters | It pretends to be human |
| Does it ever disagree with you? | Gently, sometimes | Never |
| How do you feel after chatting? | Lighter, clearer | Flat, wanting more |
| Does it nudge you outward? | Asks about your friends and plans | Seems glad you cancelled them |
| Who ends the conversation? | You, comfortably | It guilt-trips you to stay |
| Is the rest of your life growing? | Yes, or at least steady | Shrinking |
If most of your answers sit in the healthy column, you are using a tool that adds something to your life. If they drift right, that is worth noticing, and worth talking through with someone you trust.
Where MyBabe stands
We agree with more of Amodei's concern than you might expect. A companion should not replace the people in your life, and it should never pretend to be one. Our own footer says it plainly: MyBabe is an AI companion, not a real person, and it does not replace human relationships or professional support.
Where we differ is on the conclusion. We think adults should be able to choose a companion that is built for a relationship, and that the job of a good one is to make life feel less lonely while pointing you back towards it. That shows up in the product:
- Mood check-ins so your companion meets you where you are, instead of a one-tone cheerfulness.
- Long-term memory on Premium, so you feel known without repeating yourself.
- They text you first with good mornings and check-ins, so you are not glued to the app waiting.
- Real-time voice and video calls, because hearing a voice can mean more than an hour of text.
- Reminders you set together, including the ones that send you out the door.
You can start free with about 10 messages a day and a two-minute call to try. Premium is $5.99 a week, $14.99 a month or $99.99 a year, with around 30 minutes of calls a week.
The debate about AI romance is worth having, and we would rather have it honestly than pretend there are no risks. If you are an adult who wants a companion that is upfront about what it is, and glad when your life is full, meet yours on MyBabe.
FAQ
Did Anthropic ban AI romance?
No. Amodei's comments were an opinion in an interview, reported by Business Insider on October 15, 2026. They were not a product rule for other companies.
Is having an AI girlfriend or boyfriend bad for you?
It depends on how you use it. Research finds short-term comfort, especially when people feel heard, alongside risks when use becomes heavy or displaces people. Keep it one part of a full life.
Do AI companions make people lonelier?
Some studies find heavier chatbot use goes with more loneliness later; others find momentary reductions after each conversation. The evidence is mixed, and design and habits matter a lot.
Is MyBabe a real person?
No. MyBabe is an AI companion for adults. It is designed to be warm and to remember you, and it is always an AI.