"You can’t simply say a child shouldn’t use AI. This generation is growing up with it": Karandeep Anand

Named among TIME’s AI Architects 2025 alongside OpenAI’s Sam Altman and Nvidia’s Jensen Huang, Karandeep Anand is among a new wave of Indian-origin leaders shaping consumer AI. In Delhi for Synapse, a society and technology conference, Anand, the CEO of Character.AI, speaks to Sneha Bhura about why he believes AI could ultimately reduce, not deepen, screen dependence1. You’re visiting India as it wraps up its first major AI summit. As the US and China compete to win the AI race, what’s your take on India’s ambitions?It’s a fascinating time we are in. There is a mad rush to build the biggest, fastest AI models. It’s obvious that every company and every nation sees this as a form of power worth having. That’s why India, as a large emerging superpower, cannot afford to be left behind. We might be a little late from an Indian perspective, as it takes time to build the infrastructure, the technology stack, and spend the necessary capital and resources to that end. But from what I understand, the response and attendance at the AI summit were probably larger than anyone anticipated, which is a great sign of excitement and interest. While it seems the event wasn’t fully prepared to handle the logistics at that scale, you have to start somewhere. It will only get better from here.2. You stepped into the CEO’s chair at a time when Character.AI was under intense scrutiny over child safety and harmful interactions. There were chatbots called ‘Bestie Epstein’. The company later barred teens from open-ended chatbot conversations. What did that teach you about the deeper responsibilities of creating chatbots? We inherited a product that was already very successful, but at the same time, we were learning — as you said — that chatbots can be pushed in ways that challenge safety guardrails. Over the last six to seven months, I have spent a lot of time focusing on building AI responsibly. For example, we rolled out age assurance. We want to make sure that if someone says they are 18, they actually are. If you’re under 18, you can still watch and create stories with your favourite characters, but you cannot have an open-ended chat. We do a lot of work to make sure controversial characters or banned topics are automatically flagged. Users sometimes try to bypass this, like naming a banned person differently. So, we go beyond names and analyse character descriptions and images. Over the past two and a half years, we’ve built a sophisticated system to continuously monitor characters, their definitions, names and images. This isn’t just Character.AI doing it. Meta stopped its chatbots for teens, Roblox implemented age assurance and Discord did the same. What I feel really proud of is that we set the bar for what AI safety could look like.3. Govts worldwide—including India—are considering age-based restrictions on social media usage. Should AI companion platforms be regulated like social media?AI can be used to build social media but it can also be used in many other ways — as a math tutor, a language tutor, a roleplay app like ours or for AI-powered reels in Meta. It’s the use case that matters. Responsibility for guardrails cannot lie solely with regulators. Tech companies must work with regulators to explain the technology, how it can be misused and what frameworks are needed. Ultimately, it’s a combination of three stakeholders: tech companies, regulators and parents. Parents must also understand how to regulate AI usage for their children.4. You recently said in an interview that your six-year-old daughter uses Character.AI. As a parent, how do you put boundaries in place?Now that Character.AI is less accessible to teens, my daughter doesn’t use it as much, but she still interacts with it with me supervising. She also has access to ChatGPT and my Gemini account. It’s fascinating to watch someone so young use AI so naturally. That’s why you can’t simply say, “A child shouldn’t use AI.” This generation is growing up surrounded by AI, and we have to assume that, and build products accordingly. It’s a very accessible technology. Previously, when we used computers, we had to type, understand devices, operating systems and browsers. Now, a six-year-old can just speak— in Hindi, Punjabi, English — and the AI responds seamlessly. It can also adjust how it speaks to a child, differently from how a human would. That said, whether it’s for homework, storytelling or entertainment, I supervise her use. Parents need to watch which apps their kids use. Would I let my daughter use any AI-powered app from the app store? Absolutely not.5. In India, are AI companions mainly novelty and entertainment, or are they filling deeper gaps?We have roughly 20m monthly active users; 80% are outside the US, with India being a very large part of our user base. This shows that the need for AI-powered entertainment and roleplay is universal, whether it is someone in Delhi, Chandigarh or Jaipur or in New York or San Francisco. But what starts as novelty often becomes deeper usage — gaming, companionship etc.6. But dependency can be a problem for adults as well. Are there safeguards for that?Absolutely. Any conversations around violence, self-harm or sexual topics need strict moderation. We allow romantic conversations but not sexual ones. For example, a date scenario is okay, but explicit sexual dialogue is not. You can role-play fighting zombies or vampires, but not real-world shooting. Without strict moderation, a platform with AI’s power shouldn’t scale.7. What advances excite you the most?Earlier, chatbots required typing long text messages. Now, voice, natural language and multilingual bots make AI far more accessible. My mother can speak Punjabi to AI. You can even create your own characters. For example, I created a Jaspal Bhatti-like character for my parents. There’s still work to do in better supporting Indic languages. It’s an area where India can lead in innovation. Third, hyper-realistic, high-quality short videos and AI-powered games are emerging. The interesting thing is that AI can decouple people from screens. For the first time, AI allows interaction through voice and background devices, not just on a phone or computer.