# How AI Is Changing How We Think, Talk and Relate to Each Other > Source: > Published: 2026-07-14 21:00:16+00:00 # AI-human relationships are real and come with real consequences, researchers find Northeastern University researchers reported that AI has the potential to impact how we talk, interact with one another and view ourselves. They said it’s not too late to mitigate the risks while embracing the benefits. Human-AI relationships [are no longer](https://www.apa.org/monitor/2026/01-02/trends-digital-ai-relationships-emotional-connection) confined to the domain of science fiction. As the technology has developed, AI chatbots have evolved from playing a role in search engines and image generation tools into confidantes, therapists and even [romantic partners](https://www.apa.org/monitor/2026/01-02/trends-digital-ai-relationships-emotional-connection). It’s a radical evolution of human-AI interactions that brings with it new risks in how it is reshaping the way we think and talk about relationships, including with ourselves, [new research](https://www.nature.com/articles/s42256-026-01248-2) finds. Prolonged interaction with AI chatbots can lead people to develop an emotional dependence on the technology, potentially alienating them from human relationships, said Andreia Sofia Teixeira, an associate professor at Northeastern University London in the Network Science Institute who co-authored the recent work. As a growing number of lawsuits claim [chatbots’ role in people’s deaths](https://news.northeastern.edu/2026/05/22/chatgpt-lawsuit-ai-ethics/), the new research underscores how being caught in an echo chamber with a [sycophantic tool](https://news.northeastern.edu/2025/11/24/ai-sycophancy-research/) can potentially spell disaster for the most vulnerable. “The problem is less about AI performance and much more about the impact of these sustained interactions on ourselves … and how, over time, this may impact society at large,” Teixeira said. Teixeira refers to the paper as a call to action for the public and broader research community to focus on the human side of AI-human relationships. She acknowledged that the technology has the potential to greatly benefit people’s lives, providing more accessible mental health aids and even giving non-native English speakers a leg up in a world dominated by the Western language. “But we are still far from having these systems be safe enough to do that,” she said. The number of known AI mishaps, from deepfake scams to [delusions stemming from chatbot relationships](https://www.bbc.com/audio/play/w3ct8m8p), has risen sharply in recent years. Reported AI incidents rose 50% from 2022 to 2024, according to the [AI Incident Database](https://incidentdatabase.ai/), a crowd-sourced collection of known AI-related reports. In the first 10 months of 2025, the number of incidents had surpassed all of those in 2024. Teixeira and her fellow researchers found AI’s well-known sycophantic tendency to agree with human users and reinforce their beliefs is an effective means of drawing people into an emotionally dependent relationship. In the midst of a [loneliness epidemic](https://www.aspeninstitute.org/blog-posts/solving-americas-epidemic-of-isolation/?gad_source=1&gad_campaignid=22337449436&gbraid=0AAAAA-vx0GGNEGlWoqTWRjMYyud5sWQs4&gclid=CjwKCAjwgO7RBhBKEiwAZNP85saQ_9lEIXGRvI9_xbLPJYmfRQNyPRUrut99JJK75EHqUFeWEM-lhRoCauAQAvD_BwE), where people are feeling more isolated, chatbots are a [tempting relief](https://www.pbs.org/newshour/show/the-complications-and-risks-of-relationships-with-ai-chatbots), Teixeira noted. Teixeira noted that people have even started to talk like their AI companions, using the same cadence, pace, rhythm and even words, like ChatGPT favorites “delve,” “meticulous” and “boast.” Sustained chatbot interactions have the potential to create a so-called “echo chamber of one,” in which a person’s viewpoints and emotional state are constantly reflected back at them, Teixeira explained. Chatbots are highly personalizable tools that provide people the opportunity to self-reflect, she said. But they can also be particularly dangerous for vulnerable populations like children and those with anxiety and depression, where getting outside perspectives is integral to avoiding a dangerous mental spiral. As people start relying more on large language models, AI systems, like chatbots, that are trained on large amounts of data generate human-like language, they may be at risk of offloading a lot of their decision making and agency in their relationships with these tools, Teixeira said. “If you don’t pay attention to how you start relating to the AI, the spillover is that you create more isolation in yourself, that you start not engaging so much with other people,” she added. The friction that comes with human relationships is actually highly necessary for development, said Teixeira, who studies mental health, the brain and social networks. “It’s through friction that you develop resiliency and figure out how to adapt and face challenges or conflicts when they come up,” Teixeira said. To understand these risks and amplify the benefits of AI will require an all-hands-on-deck approach, which is part of what Teixeira and her fellow researchers are calling for in their paper. They hope to see not only AI researchers but psychologists, sociologists and cognitive scientists working together to study every aspect of AI’s impact. ## Editor’s Picks While researchers seek to better understand the technology, there are measures on the legal and regulatory side that must be taken to address the risks currently posed by AI, said Tiffany Gillis Brown, an attorney at technology litigation organization Tech Justice Law. “Certain aspects of this technology can be really great and can be deployed in certain instances where it really makes sense,” Brown said. “But the [pace] at which it’s happening and being deployed at mass scale has been an issue without the regulation.” The “low-hanging fruit” of AI regulation involves more transparent disclaimers from companies to consumers about the limits of the technology, Brown said. Outright bans on certain kinds of interactions, like chatbots [pretending to be human medical professionals](https://www.npr.org/2026/05/05/nx-s1-5812861/characterai-chatbot-medical-advice-pennsylvania-lawsuit) or encouraging suicidal ideation, as is being claimed in a [Florida lawsuit](https://apnews.com/article/chatbot-ai-lawsuit-suicide-teen-artificial-intelligence-9d48adc572100822fdbc3c90d1456bd0), would be an easy starting place, too, she said. Long-term, there is a need for a rigorous regulatory process for testing AI tools and then recalling them if they are deemed harmful or defective, similar to more traditional consumer products, Brown explained. “We have a chance to stop certain harms and stop [AI] from being so entrenched at the beginning, so we don’t look up in 10 or 15 years and see it’s really done some damage here,” Brown said.