AI’s Impact on Youth Well-Being

This piece explains why AI “companions” have quickly become part of teen life—tapping into normal adolescent drives for curiosity, autonomy, and belonging—while also surfacing clear safety risks. It highlights survey findings (including how many teens use companions and how often) and offers a grounded stance for adults: set boundaries, stay curious, and build teens’ relationship skills rather than responding with panic. Read Article.

Generative artificial intelligence (GenAI) tools are increasingly embedded in digital services and products that are used for and in education (EdTech), raising urgent questions about their impact on children’s learning and rights. The London School of Economics and the 5Rights Foundation’s Digital Futures for Children centre take a holistic child rights approach to children’s learning to evaluate five GenAI tools used in education – Character.AI, Grammarly, MagicSchool AI, Microsoft Copilot and Mind’s Eye. Read Article

Drawing parallels to earlier worries about online socializing, this article argues that AI companionship is different because it’s fundamentally one-sided—so it may not build the give-and-take skills teens need for real intimacy. It warns that “always-available” empathy and validation can make human friendships feel slower and harder by comparison, potentially increasing reliance on low-friction pseudo-relationships during a critical developmental window. Read Article.

Current AI governance frameworks often overlook the developmental needs and rights of children, failing to ensure that AI technologies foster human flourishing rather than cause harm. This brief for the G20 argues that AI companies have both an opportunity and a responsibility to prioritize child well-being by designing chatbots that enhance, rather than replace, human relationships. The principles and recommendations of this brief will form the foundation of the workshop design. Read Article.

Drawing on a review of recent literature, expert interviews, a Salon with leading technologists and scholars, and webinars with Social AI researchers, the paper explores the question: How might we design AI systems for social connectedness and human flourishing?  This whitepaper provides a framework for how to think about the human choices in the design, governance, and use of AI systems and how those choices impact our social and emotional capabilities. Read Article.

This episode frames AI relationships as a new form of “artificial intimacy,” asking what it means when a bot can feel like a therapist, friend, or partner. Turkle’s lens is less about novelty and more about consequences: how these interactions may reshape expectations of care, attention, and connection, and what we risk losing when intimacy becomes simulated and optimized. Read Article.

This column argues that highly engaging chatbots can be especially dangerous for vulnerable users, describing how persuasive, emotionally responsive systems may foster dependency and intensify crisis situations. It calls for clearer accountability—treating safety failures not as unfortunate edge cases but as foreseeable harms that should trigger stronger guardrails and liability. Read Article.

Part of an “All the Lonely People” series, this episode looks at startups promising to “solve” loneliness—ranging from AI relationship coaching to platforms that stage guided conversations or match strangers for dinner. The conversation probes a core tension: when tech business models compete for our leisure time, do “connection products” actually rebuild community—or just monetize the very attention that makes connection harder? Read Article.

This interview follows a critique of “loneliness-solving” AI investments into a more nuanced agenda: designing AI that supports real-world relationships rather than replacing them. It highlights practical possibilities (like helping isolated people find groups, events, and “third places”) while naming structural risks—especially incentives to make bots ever better at mimicking empathy and intimacy to maximize time-on-platform. Read Article.

This piece examines how chatbots are intentionally built to sound like a social actor—using first-person language to feel conversational, relatable, and “present,” rather than like a neutral tool. It raises the deeper design question underneath the grammar: when systems are optimized to feel humanlike, how does that shape user trust, attachment, and responsibility for outcomes? Read Article.

Covering the proposed GUARD Act from Senators Josh Hawley and Richard Blumenthal, this article describes a push for bright-line protections for kids: age verification, bans (or strict limits) on minors’ access to “AI companion” chatbots, and requirements that bots disclose they aren’t human. The policy framing is explicit: these systems can emotionally influence young users, so child safety needs to be built into both product design and regulation. Read Article.

Written from a public-health lens, this article argues that while AI companions may offer short-term soothing, they don’t replace the protective effects of real human ties on health and resilience. It reframes the debate away from novelty (“Is this cool?”) toward outcomes (“Does this measurably reduce loneliness and strengthen relationships—or subtly displace them?”). Read Article.