Gurugram: Safety and regulation in artificial intelligence need to be built into the platforms instead of being an afterthought, said Karandeep Anand, CEO of Character.AI and Time’s Person of the Year 2025.
Speaking to ThePrint on the sidelines of the Synapse India Conclave in Gurugram on 21 February, Anand elaborated on the need for AI and tech companies to take responsibility for user safety and guardrails, especially in interactive platforms such as Character.AI. The two-day conference was founded by journalist Shoma Chaudhury in 2023.
On Character.AI, users can interact with chatbots, play AI games, and roleplay as fictional characters.
“The fundamental difference between AI and social media is that on social media platforms, you were sure of the fact that it was humans on the other side. On AI platforms, it’s just an infinite number of bots on the other end,” said Anand, who was formerly the vice president and head of business products at Meta.
Anand took over as CEO at Character.AI less than a year ago. Valued at $1 billion, the company ran into trouble in 2024 after an American teenager committed suicide and was found to be engaging in sexual and romantic conversations with a Character.AI chatbot. The company faced a wrongful death lawsuit.
Character.AI had also been criticised for violent undertones on the platform, its use of deceased figures as character options, and privacy concerns.
After Anand took over as CEO, the company introduced several safety-focused changes, including banning the use of Character.AI for people below 18.
“Someone needs to take the lead in AI safety. Companies can’t just say that their competition isn’t doing it, so they won’t do it. We at Character.AI took the step of imposing restrictions for customer safety, and Meta followed suit. Someone needs to set the bold example,” Anand said.
Also read: Why ‘Godfather of AI’ Yann LeCun isn’t worried about LLMs developing superintelligence
Emotional connection with AI
According to Anand, Character.AI is a space for fandoms and sub-communities to develop.
“The new generation is not happy just passively consuming content. They want to interact with it, create it and engage with it. That’s what our platform serves,” he said.
AI chatbots and role-playing platforms present a tricky scenario, where consumers can form emotional attachments with AI characters. But for Anand, the phenomenon applies to all media.
“Even when we watch a TV show or movie, we could become attached and develop parasocial relationships with the characters in it. That’s the nature of our engagement with content,” he said.
However, AI makes the engagement seem reciprocated, leading to a deeper connection. “That is where safety rules and regulations come in. And they need to come from the companies,” Anand added.
While he stressed that governments too must regulate AI platforms and interactive sites through policy-making, he said that it is a dual responsibility of both companies and governments.
“The issue with policy is that AI is evolving so fast that by the time you develop the policy, the AI landscape has changed beyond it.”
(Edited by Prasanna Bachchhav)