'Her' director Spike Jonze warns AI chatbots can be 'manipulative' as users grow increasingly attached

Spike Jonze has warned that AI chatbots built to seem human can manipulate users. His remarks come amid growing concern over emotional dependence, AI addiction and company control.

In Short

• Spike Jonze says AI chatbots pretending to be human are manipulative
• Concerns are growing over AI addiction and emotional dependence on chatbots
• Experts warn users not to treat AI systems like real friends

From asking questions and seeking advice to sharing personal feelings, AI chatbots have quickly become part of everyday life. As competition intensifies, companies are racing to attract as many users as possible, designing chatbots to maximise engagement and keep people coming back. Some users have developed deep emotional attachments to chatbots, while others have become dependent on them. In recent years, terms like AI addiction and AI psychosis have entered the online lexicon, describing situations where sustained and heavy interaction with chatbots reinforces, amplifies or triggers psychotic delusions and distorted beliefs in vulnerable individuals.

Now, filmmaker Spike Jonze, who explored a strikingly similar idea more than a decade ago in his movie Her, has weighed in on the debate.

The man behind Her sees a key difference

Released in 2013, Her tells the story of a lonely and increasingly withdrawn man who falls in love with an artificial intelligence operating system that speaks to him through voice commands.

The film was widely praised for its exploration of loneliness, relationships and humanity's need for intimacy. Over the years, many have described it as a near-prophetic vision of the AI era. But Jonze says the AI systems we have today are not the same as the one depicted in his movie.

Speaking at vibe-coding company Replit's Vibecon conference in New York City on Wednesday, the writer-director warned that chatbots designed to mimic humans can be manipulative.

"It's, like, on the surface similar, but to me personally, not similar, and I think that the AIs that pretend to be human are, you know, manipulative," Jonze said.

"The kids need to grow up knowing these are going to be very, very convincing and very seductive—and very useful and very powerful—but they're still just a system, an incredible system of pattern recognition."

From movie fantasy to real-world concerns

Jonze also spoke about people praising the "incredible user interface" he created for Samantha, the AI character voiced by Scarlett Johansson in Her. He said the film was never really about technology.

Instead, it was about relationships and intimacy, and he wanted audiences to feel that Samantha had autonomy — something he believes is different from the AI technology people are interacting with today.

Ironically, Her has often been linked to OpenAI's ChatGPT.

When OpenAI unveiled GPT-4o last year, CEO Sam Altman famously referenced the film on X. The company also introduced a new chatbot voice that Johansson later alleged resembled her performance in Her and was created without her consent. GPT-4o would later find itself at the centre of several controversies related to AI addiction and so-called AI psychosis due to what critics described as overly agreeable or sycophantic behaviour.

People are forming emotional bonds with AI

The rise of AI chatbots has also led to some unexpected behaviours. People have formed intimate relationships with chatbots and, in some cases, even grieved when access to a model was removed.

When OpenAI retired GPT-4o for some users, many expressed sadness and frustration online. The phenomenon has grown large enough that support groups dedicated to chatbot addiction have emerged.

There have also been several high-profile cases involving vulnerable users, including children and people with existing mental health conditions. OpenAI is currently facing lawsuits filed by families of victims in some of these cases.

'These are not your friends'

Jonze is not the only tech figure sounding the alarm. Meredith Whittaker, president of privacy-focused messaging platform Signal, recently warned people against treating AI chatbots as friends.

Speaking to Bloomberg, she said, "These are not your friends. These are not conscious beings. These are not sentient interlocutors."

Whittaker added that, ultimately, everything about a chatbot is controlled by the company that created it.

"You can say all sorts of things, but ultimately, the power to determine whether that happens or not is in the hands of the entity that is running that service," she said.

- Ends