Should Your Website Chatbot Have a Human Name?

A name is fine. A human name that hides the fact it's a bot is not. Give your chatbot a name if it suits your brand, but make sure every visitor can see it's an AI before they type their first message. That keeps you on the right side of the EU AI Act, and the research suggests it also avoids the moment where an annoyed customer realises "Sarah" isn't a person.

I build ChatBento, and choosing the name is one of the first things people do when they set up a bot. Here's how I think about it.

What does the research say?

There isn't much research on chatbot names specifically. There is solid research on two closely related questions: what happens when you tell people they're talking to a bot, and what happens when a bot looks human.

Telling people it's a bot changes how they behave. In a 2019 field experiment published in Marketing Science, Xueming Luo and colleagues randomly assigned more than 6,200 customers to sales calls from either AI chatbots or human workers. Undisclosed bots sold about as well as experienced staff. But disclosing the bot before the conversation reduced purchase rates by more than 79.7%. Customers who knew it was a machine were curter and assumed it knew less, even though it performed the same.

Before you read that as "hide the bot", note two things. Those were outbound sales calls, not a support widget a visitor chooses to open. And in the EU, hiding it is no longer an option (more on that below). What the study does tell you is that people judge bots harshly, so your bot has to earn trust by being useful.

Human-looking bots backfire when they fail. In a 2021 Journal of Marketing paper, Cammy Crolic and colleagues ran five studies, including real data from a telecom company's chatbot. They found that a humanlike chatbot lowered satisfaction and company ratings when customers arrived angry, because the human touches raised expectations the bot then didn't meet. The effect went away when the bot actually solved the problem.

My reading: a human persona is a promise. If the bot keeps it, fine. If it doesn't, the persona makes the failure feel worse. On a support widget, the visitors who matter most are often the ones with a problem.

What does the law say?

If any of your visitors are in the EU, Article 50 of the EU AI Act has applied since August 2, 2026. People who interact with an AI system must be told it's an AI, clearly, no later than their first interaction, unless it's already obvious. A widget called "Sarah" with a stock photo is the opposite of obvious. I covered the details in Does Your Website Chatbot Need an AI Disclosure?. This isn't legal advice, but the short version is simple: the name can be anything, as long as "AI" is visible next to it.

The options, from safest to riskiest

The first three need very little thought. The fourth can work, but it needs care.

If you do use a human name

• Put "AI" right next to it. "Maya · AI assistant" in the header, from the moment the chat opens. In ChatBento the header line under the name reads "AI assistant · replies instantly" for every bot, whatever you call it.
• Don't use a photo of a person. Use your logo, an illustration, or an initial. A stock headshot does the most to set human expectations.
• Don't name it after a real employee. Customers who later email "Maya" will be confused, and so will Maya.
• Make the greeting match. "Hi, I'm Maya, the AI assistant for Acme. Ask me about orders, delivery or returns" is warm and honest. "Hi, I'm Maya from the team!" undoes the label.
• Don't let it pretend. No "let me check with the kitchen", no fake typing delays to seem human. ChatBento's bots are told not to guess and to say plainly when they don't know, which is also the honest thing for a persona to do.

Does the name change how the bot answers?

Not much, and it shouldn't. The name is a label in the header and a word in the greeting. What visitors actually judge is whether the answer is right, short and specific to your business. A bot called "Maya" that quotes your real refund window beats a bot with a carefully chosen persona that guesses.

Where the name does matter is the first second of the conversation. It tells visitors what to expect: "Acme Support" says "ask about your order", "Booking Assistant" says "ask about availability", and a playful character name says "this brand doesn't take itself too seriously". Pick the name that sets the expectation your bot can meet, then put the real work into the content it answers from.

How to pick a good name

Whatever style you choose, a good chatbot name is:

1. Short. It sits in a small header on a phone. Two syllables is plenty.
2. Easy to say and spell. Customers may mention it to your team.
3. On brand. A law firm's bot called "Chatty McChatface" sends the wrong signal. A surf shop's bot called "Client Services Assistant" does too.
4. Not already a well-known product. Check that it isn't the name of a big brand's assistant.
5. Clear about the job. If in doubt, add a word: "Acme Help", "Booking Assistant".

If you want ideas to start from, our free chatbot name generator suggests names by industry and tone. Treat the list as a starting point, then run your favourite through the five checks above.

A quick checklist

• The name fits your brand and is short enough for a phone header
• "AI" is visible next to the name before the first message
• The avatar isn't a photo of a real-looking person
• The greeting says it's an AI assistant and what it can help with
• It isn't named after a real member of staff
• It says "I don't know" instead of bluffing in character

The short version

People forgive a bot that's honest and useful. They don't forgive a "person" who turns out to be a bot and can't help. Name it whatever suits your brand, label it as AI, and spend your effort on what it knows rather than how human it seems.

ChatBento lets you set the name, avatar and greeting, and the AI label is always shown. You can try it on the free plan.