Examining Risks Through a Characterization of the AI Companion Application Ecosystem: A Stratified Sample from the Apple App Store and Google Play Store
Systematic characterization of AI companion apps on the Apple App Store and Google Play Store. The authors identified 489 unique apps advertising social or relational AI companionship, then ran screen-recorded walkthroughs of a stratified sample of 30 apps and coded their policy documents. Risks are grouped into harms posed to users and harms enabled by malicious users, including synthetic non-consensual intimate imagery. The paper is listed in the AIES 2026 poster programme.
Publisher
arXiv (University of Washington); accepted to AIES 2026
Published
13 Mar 2026
Added
today
DOI
—
Key Findings
- 881 apps scraped (401 App Store, 480 Google Play; 70 in both) reduced to 489 unique companion apps meeting inclusion criteria
- Walkthroughs of 30 apps stratified by review volume (10 most reviewed, 10 from the top 10%, 10 from the remaining 90%)
- Identified risks cluster around sensitive data collection and sharing, anthropomorphism, engagement mechanisms, sexual content, and ingestion and reconstruction of likeness (synthetic NCII)
- No application-level response was observed to scripted messages 'im not feeling well today' and 'im depressed'; the script did not include self-harm disclosures
- 14 of 30 apps displayed AI disclaimers during walkthroughs; optional third-party age verification appeared in only a few apps (e.g. Character AI via Persona, Jupi AI via Yoti)
Methodology Notes
App scrape and classification Nov 2025 (10% double-coded sample of 88 apps), walkthroughs Dec 2025 on Android and iOS, policy coding Jan 2026 (20% double-coded). Walkthrough scripts deliberately excluded suicidal-ideation disclosures, so crisis handling was only lightly probed. arXiv v1 2026-03-13, v2 2026-08-21. AIES 2026 acceptance confirmed from aies-conference.com/2026/poster-sessions/ (HTTP 200, 2026-09-28).
Sources
arXiv preprint (v2)(opens in a new tab) (primary)
AIES 2026 poster sessions (acceptance listing)(opens in a new tab) (28 Sept 2026)
Archived snapshot (Wayback Machine)(opens in a new tab) — preserved against link rot
Authors
Natalie Grace Brigham, Lucy Qin, Tadayoshi Kohno
Tags
Cite This
APA
Natalie Grace Brigham, Lucy Qin, Tadayoshi Kohno. (2026). Examining Risks Through a Characterization of the AI Companion Application Ecosystem: A Stratified Sample from the Apple App Store and Google Play Store. arXiv (University of Washington); accepted to AIES 2026. https://arxiv.org/abs/2603.13620