Lensa AI Magic Avatars undresses journalists without permission
Occurred: December 2022
Report incident 🔥 | Improve page 💁 | Access database 🔢
Journalists were easily able to generate near-realistic adolescent and sexualised images of themselves using Lensa's Magic Avatars.
Journalists at Technology Review, TechCrunch, and Insider recorded how they were easily able to generate near-realistic adolescent and sexualised images of themselves.
The findings prompted concerns about the effectiveness of the app's safety guardrails.
Lensa’s terms of service oblige users to submit only appropriate content, including 'no nudes'.
Lensa owner Prisma Labs CEO Andrey Usoltsev told TechCrunch that such behaviour 'can only happen if the AI is intentionally provoked' into creating unsafe content and blamed users for breaching its terms of use.
System 🤖
Operator: Melissa Heikkilä; Zoe Sottile
Developer: Prisma Labs
Country: UK; USA
Sector: Media/entertainment/sports/arts
Purpose: Generate avatars
Technology: Neural network; Deep learning; Machine learning
Issue: Stereotyping
News, commentary, analysis 🗞️
https://www.technologyreview.com/2022/12/13/1064841/download-ai-objectification-sbf-charged/
https://www.businessinsider.com/lensa-ai-raises-serious-concerns-sexualization-art-theft-data-2023-1
https://lifestyle.livemint.com/smart-living/innovation/why-being-air-brushed-by-lensa-ai-made-me-anxious-111671865909643.html
https://www.entrepreneur.com/business-news/what-is-lensa-ai-app-and-is-it-dangerous-for-your/441148
https://edition.cnn.com/style/article/lensa-ai-app-art-explainer-trnd/index.html
Page info
Type: Incident
Published: December 2023