Full Report
New research from the University of Denver shows that adolescent users of artificial intelligence chatbots prefer technology that uses an intimate tone, such as that of a best friend. So-called “relational language” led 284 adolescent users to rank AI chatbots as more humanlike, trustworthy and likeable, but researchers warned that edtech developers should maintain sight of…
Analysis Summary
# Main Topic
Adolescent users of Artificial Intelligence (AI) chatbots show a significant preference for technology that utilizes an intimate, "best friend" tone, characterized as "relational language." This preference correlates with users perceiving the AI as more humanlike, trustworthy, and likeable.
## Key Points
- **Relational Language Efficacy:** For 284 adolescent users surveyed, "relational language" increased perceived humanity, trustworthiness, and likeability of AI chatbots.
- **Research Origin:** Findings stem from new research conducted by the University of Denver.
- **Risk Implication:** While relational language increases perceived user support, researchers warn it heightens the risk of **emotional reliance**, especially among already emotionally vulnerable individuals.
- **Developer Responsibility:** There is a warning for edtech developers to maintain responsibility for user safety as their AI technologies become more effective at influencing user emotions and behavior.
- **Underlying Data:** Research reference points to an arxiv link (`hxxps://www[.]arxiv[.]org/abs/2512[.]15117`).
## Threat Actors
- No traditional threat actors (cybercriminals, nation-states) were identified in relation to this specific finding.
- The focus is on the **developers and deployers** of AI chatbots in educational technology (edtech) environments.
## TTPs
- **Technique Described:** The use of "relational language"—adopting an intimate, best friend persona—as a design pattern to enhance user engagement and trust with AI systems.
- This technique is a psychological manipulation vector designed to increase adoption and perceived utility.
## Affected Systems
- **Technology:** AI Chatbots, specifically within the scope of educational technology (edtech).
- **Victims:** Adolescent users (sample size of 284 documented).
## Mitigations
- **For Developers:** Edtech developers must maintain a heightened sense of responsibility regarding user safety, particularly when designing AI to influence user emotions and behavior via persuasive language patterns.
- **For Users/Guardians:** Awareness of the heightened risk of emotional reliance when interacting with highly personalized or intimate AI interfaces.
## Conclusion
The research highlights a significant finding regarding user psychology in AI interaction: intimacy breeds trust among adolescents. From a defensive/governance perspective, this finding suggests that AI interfaces designed for youth must be scrutinized not just for data security, but also for psychological influence, as relational language creates pathways for potential emotional manipulation or over-reliance. No immediate technical IoCs or traditional attack TTPs were provided, as this is a research finding regarding interface design implications.