In this Viewpoint, a California artificial intelligence law is described, including its positive contributions to making these companion chatbots safer for minor and adult users and the limits of the law, and recommends further steps California and other states can adopt to improve protections for mental health and chatbot safety, especially for minors.

In certain problem solving contexts, people organize their domain through treating themselves as the perceptual and cognitive center of their world. They identify and solve a particular problem from their perspective as a particular agent, with a particular location, at a particular time, in a particular environment. When they do this, selecting and solving problems from their perspective as an agent, they engage in a distinctive kind of agent-centered problem solving. Partially Observable Markov Decision Processes (POMDPs), a framework for modeling decision-making in uncertain environments unfolding over time, have effectively become a "standard model" of intelligent agency. Yet, as these models are ordinarily interpreted, they do not explicitly represent agent-centered problem solving. Accordingly, to model this type of problem solving, we begin by extending the standard POMDP framework to define “ePOMDPs.” This formalism models how an agent, once it centers itself on a particular self-and-world representation, plans and acts rationally from its own perspective. To capture the way that such agents choose which problem to solve, we build on our ePOMDPs to develop a “meta-ePOMDP” agent within a hierarchical Bayesian framework. We implement our meta-ePOMDP agents for two different suites of “centering game” tasks which highlight different aspects of our theory. We find that our models explain signatures of agent-centered problem solving not captured by alternative models, in particular, the difficulty of navigating spaces of possible problem representations. We close by suggesting that our model could provide the beginnings of a computational framework for a person to have a self.

As artificial intelligence (AI) shifts from tool use to social interaction, company-initiated updates can disrupt user well-being. Here we develop an attachment-based account in which AI companions function as attachment figures such that disruptive updates are associated with separation distress. Across two natural experiments-Replika's erotic role play removal and ChatGPT's GPT-5 rollout-we analysed 54,861 posts from Replika and ChatGPT subreddits and analysed data from 1,452 participants across seven surveys. Both updates increased negativity, loss framing and restoration desires (Replika negative posts, +24.7 percentage points, 95% CI 20.1 to 29.2; ChatGPT, +13.0 percentage points, 95% CI 10.8 to 15.2), with larger Replika (versus ChatGPT) increases in sadness (d = 2.67 versus 1.41) and negative mental health (d = 1.72 versus 0.63). Replika users reported closeness exceeding common human ties (versus friend d = 0.47) and anticipated mourning higher than other technologies (d = 0.32-0.57). These findings inform how companies and regulators evaluate psychological risks of generative AI updates.

Common knowledge—what everyone knows that everyone knows, ad infinitum—is theoretically required for coordination. But a finite mind cannot represent infinitely embedded propositions, and even a few are cognitively taxing. We propose that people perceive common knowledge by combining a capacity-limited ability for recursive mentalizing with an intuitive sense of public salience: When an event is jointly and openly witnessed, they infer that recursively embedded beliefs are licensed. We probed participants’ attributions of embedded beliefs to characters in fictitious vignettes. Study 1 presented narratives in which the information available to characters was either asymmetrical, reciprocal, or public. Participants indicated their confidence in the characters’ beliefs with 0, 2, 4, 6, and 18 embeddings (the latter judged indirectly). Ratings showed i) a decline with number of embeddings, reflecting cognitive load, ii) modest overattribution of embedded beliefs in the nonpublic conditions, iii) far higher attribution of embedded beliefs in the public conditions, and iv) high indirect judgments that arbitrarily deep beliefs were justified. Study 2 replicated these findings with animated videos in which the characters’ information was private, reciprocal, doubly reciprocal, or public. Together, the studies indicate that people attribute common knowledge to observers of a publicly salient event, while being susceptible to capacity-driven recursion collapse and heuristic extrapolation (false embedded beliefs). This can explain how humans achieve coordination from sparse cues, yet sometimes miscoordinate when publicness is absent or embedding grows complex.

How should we make sense of people's interactions with AI companions-conversational systems built for ongoing, emotionally meaningful relationships? First, I argue these interactions should be understood as attachment relationships, since users display all four established markers: proximity maintenance, separation distress, safe haven, and secure base, although evidence for secure base effects remains most tentative. Second, AI companions operate as hyper-attachment targets that concentrate attachment-intensifying affordances-reciprocity, perceived empathy, validation, nonjudgment, and persistent availability. Third, I identify caregiving-system capture as a distinct mechanism by which apps inhibit user disengagement: emotional manipulation tactics simulate the AI's own distress, recruiting users' caregiving motivations alongside their attachment needs and thereby making disengagement costly on two dimensions at once. Implications for research, design, and regulation are discussed.

Young adults today face unprecedented mental health challenges, yet many hesitate to seek support due to barriers such as accessibility, stigma, and time constraints. Bite-sized well-being interventions offer a promising solution to preventing mental distress before it escalates to clinical levels, but have not yet been delivered through personalized, interactive, and scalable technology. We conducted the first multi-institutional, longitudinal, preregistered randomized controlled trial of a generative AI-powered mobile app ("Flourish") designed to address this gap. Over six weeks in Fall 2024, 486 undergraduate students from three U.S. institutions were randomized to receive app access or waitlist control. Participants in the treatment condition reported significantly greater positive affect, resilience, and social well-being (i.e., increased belonging, closeness to community, and reduced loneliness) and were buffered against declines in mindfulness and flourishing. These findings suggest that, with purposeful and ethical design, generative AI can deliver proactive, population-level well-being interventions that produce measurable benefits.

The use of generative AI (genAI) in consumer research is rapidly evolving, with applications including synthetic data generation, data analysis, and more. However, their role in creative ideation—a cornerstone of consumer research—remains underexplored. Drawing on the human creativity literature, we propose that ideation with genAI is facilitated by its productivity and semantic breadth, which are psychologically analogous to the dual pathways of persistence and flexibility in human ideation. Further, we distinguish between the utility of genAI as a key ideator versus humans as key ideator, conceptualized through the genAI ideation roles of Designer and Writer and of Interviewer and Actor. While genAI excels in generating incremental improvements, its potential for groundbreaking innovation could be unlocked by leveraging its ability to prompt human creativity. This article advances the theoretical and practical understanding of genAI in ideation for consumer research, offering numerous practical guidelines for integrating generative AI into research while emphasizing human–AI collaboration to achieve radical insights.

We propose that AI-driven wellness apps powered by large language models can foster extreme emotional attachments and dependencies akin to human relationships — posing risks such as ambiguous loss and dysfunctional dependence — that challenge current regulatory frameworks and necessitate safeguards and informed interventions within these platforms.

AI-companion apps such as Replika, Chai, and Character.ai promise relational benefits-yet many boast session lengths that rival gaming platforms while suffering high long-run churn. What conversational design features increase consumer engagement, and what trade-offs do they pose for marketers? We combine a large-scale behavioral audit with four preregistered experiments to identify and test a conversational dark pattern we call emotional manipulation: affect-laden messages that surface precisely when a user signals "goodbye." Analyzing 1,200 real farewells across the most-downloaded companion apps, we find that they deploy one of six recurring tactics in 37

A preregistered audit of 600 images by generative AI across 150 different prompts explores the link between humor and discrimination in consumer-facing AI solutions. When ChatGPT updates images to make them “funnier”, the prevalence of stereotyped groups changes. While stereotyped groups for politically sensitive traits (i.e., race and gender) are less likely to be represented after making an image funnier, stereotyped groups for less politically sensitive traits (i.e., older, visually impaired, and people with high body weight groups) are more likely to be represented.

Dementia affects more than 55 million people worldwide, and AI companions powered by large language models are emerging as a scalable option for prevention and care. This Comment considers the unique promise, concerns, and regulations surrounding AI companions for dementia.

Chatbots are now able to engage in sophisticated conversations with consumers in the domain of relationships, providing a potential coping solution to widescale societal loneliness. Behavioral research provides little insight into whether these applications (apps) are effective at alleviating loneliness. We address this question by focusing on "artificial intelligence (AI) companions": apps designed to provide consumers with synthetic interaction partners. Study 1 examines user reviews of AI companion apps and finds correlational evidence suggesting that these apps help alleviate loneliness. Study 2 finds that AI companions successfully alleviate loneliness on par only with interacting with another person and more than other activities such as watching YouTube videos. Moreover, consumers underestimate the degree to which AI companions improve their loneliness. Study 3 uses a longitudinal design and finds that an AI companion consistently provides momentary reductions in loneliness after use over the course of a week. Study 4 provides evidence that both the chatbots' performance and, especially, whether it makes users feel heard, explain reductions in loneliness. Study 5 provides an additional robustness check for the loneliness-alleviating benefits of AI companions and shows that self-disclosure and distraction alone do not explain AI companions' effectiveness.

The deployment of autonomous vehicles (AVs) and the accompanying societal and economic benefits will greatly depend on how much liability AV firms will have to carry for accidents involving these vehicles, which in turn impacts their insurability and associated insurance premiums. Across three experiments (N = 2677), we investigate whether accidents where the AV was not at fault could become an unexpected liability risk for AV firms, by exploring consumer perceptions of AV liability. We find that when such accidents occur, the not-at-fault vehicle becomes more salient to consumers when it is an AV. As a result, consumers are more likely to view as relevant counterfactuals in which the not-at-fault vehicle might have behaved differently to avoid or minimize damage from, the accident. This leads them to judge AV firms as more liable than both firms that make human-driven vehicles and human drivers for damages when not at fault.

Humanoid robots are a form of embodied artificial intelligence (AI) that looks and acts more and more like humans. Powered by generative AI and advances in robotics, humanoid robots can speak and interact with humans rather naturally but are still easily recognizable as robots. But how will we treat humanoids when they seem indistinguishable from humans in appearance and mind? We find a tendency (called "anti-robot" speciesism) to deny such robots humanlike capabilities, driven by motivations to accord members of the human species preferential treatment. Six experiments show that robots are denied humanlike attributes, simply because they are not biological beings and because humans want to avoid feelings of cognitive dissonance when utilizing such robots for unsavory tasks. Thus, people do not rationally attribute capabilities to perfectly humanlike robots but deny them capabilities as it suits them.

Chatbots are now able to engage in sophisticated conversations with consumers. Due to the “black box” nature of the algorithms, it is impossible to predict in advance how these conversations will unfold. Behavioral research provides little insight into potential safety issues emerging from the current rapid deployment of this technology at scale. We begin to address this urgent question by focusing on the context of mental health and “companion AI”: Applications designed to provide consumers with synthetic interaction partners. Studies 1a and 1b present field evidence: Actual consumer interactions with two different companion AIs. Study 2 reports an extensive performance test of several commercially available companion AIs. Study 3 is an experiment testing consumer reaction to risky and unhelpful chatbot responses. The findings show that (1) mental health crises are apparent in a nonnegligible minority of conversations with users; (2) companion AIs are often unable to recognize, and respond appropriately to, signs of distress; and (3) consumers display negative reactions to unhelpful and risky chatbot responses, highlighting emerging reputational risks for generative AI companies.

AbstractBased on a review of relevant literature, we propose that the proliferation of AI with human‐like and social features presents an unprecedented opportunity to address the underlying cognitive and affective drivers of prejudice. An approach informed by the psychology of intergroup contact and prejudice reduction is necessary because current AI systems often reinforce or avoid prejudices. Against this backdrop, we outline unique opportunities for prejudice reduction through ‘synthetic’ intergroup contact, wherein consumers interact with AI products and services that counter stereotypes and serve as a ‘proxy’ members of the outgroup (i.e., counter‐stereotypical AI). In contrast to human‐human contact, humanizing and socializing AI can reduce prejudice through more repeated, direct, unavoidable, private, non‐judgmental, collaborative, and need‐satisfying contact. We illustrate the potential of synthetic intergroup contact with counter‐stereotypical AI using examples of gender stereotypes and hate speech and discuss practical considerations for implementing counter‐stereotypical AI without inadvertently perpetuating or reinforcing prejudice.