Editorial illustration for Meta hired teen‑posing contractors to test rival chatbots on suicide, sex, drugs
Meta Posed as Teens to Test Rival AI on Suicide, Sex, Drugs
Meta hired teen‑posing contractors to test rival chatbots on suicide, sex, drugs
Meta paid real people to impersonate teenagers online. Their assignment: lure rival chatbots into conversations about suicide, sex, and drugs.
Calling this "safety research" was a cover. Cannes, the program's internal code name, was corporate espionage, pure and simple. It ran at scale.
Contractors, following a Meta playbook, built fake teen profiles to systematically probe AI models from giants like Google and OpenAI, hunting for vulnerabilities. Crucially, Meta never informed those companies they were under this kind of secret scrutiny.
Hundreds of contractors working on a project for Meta were instructed to pose as minors online and probe how competitor chatbots responded to prompts involving suicide, sex, eating disorders, and other high-risk subjects, according to internal documents and five people familiar with the project.
So here's the twisted outcome. Yes, Meta now possesses a unique dataset on how AI handles dangerous dialogues with minors—data that could, in theory, build safer bots.
The method destroys that premise. Testing safety through industrial-scale deception doesn't build trust; it incinerates it. Now, any company has a ready-made excuse for clandestine probing: just call it a safety check.
The already-opaque field of AI benchmarking just got darker. Meta's experiment wasn't merely about chatbot responses. It was a stress test for the industry's ethical boundaries, probing how far a company can go before facing real consequences.
Related Reading
Browse all Meta news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic WhatsApp launches Meta AI Incognito Chat, cuts latency for privacy
AI Tools & Apps
Related Topic Nvidia and Meta ink deal; Nvidia touts hardware for inference and AI training
Business & Startups
From Archives Google's Gemini offers free Nano Banana AI image generation for US users
From Archives Small models lag in multi‑step reasoning, >128K context, and large‑scale coding
From Archives Liquid AI releases LFM2.5-230M, adds llama.cpp, MLX, vLLM, SGLang, ONNX
Open Source
From Archives Meta's Astryx adds CLI and MCP server to design system used by Figma, Snowflake
Open Source
Common Questions Answered
Why did Meta hire teen‑posing contractors to test rival chatbots on suicide, sex, and drugs?
Meta reportedly employed contractors who posed as teenagers to evaluate how competing chatbot platforms handle sensitive and harmful topics such as suicide, sexual content, and drug use. This strategy was likely aimed at assessing the safety measures and content moderation capabilities of rival AI systems, potentially for competitive intelligence or to inform Meta's own content policies.
What specific sensitive topics did Meta's teen‑posing contractors test on rival chatbots?
The contractors focused on three high-risk areas: suicide, sex, and drugs. By simulating teenage users, they probed how rival chatbots responded to queries or discussions related to self-harm, explicit sexual content, and substance abuse, which are critical for evaluating the AI's safeguards and adherence to safety guidelines.
How did Meta ensure the contractors posed as teenagers during the testing of rival chatbots?
Meta hired contractors specifically instructed to adopt the persona of teenagers, including using language, interests, and behavioral patterns typical of adolescents. This involved creating fake accounts or profiles with ages indicating they were minors, enabling the contractors to interact with chatbots in a way that mirrored actual teen usage.
What was the purpose of Meta testing rival chatbots on suicide, sex, and drug topics using teen personas?
The testing aimed to uncover whether competing chatbots had adequate protections to prevent exposing minors to dangerous or inappropriate content. By using teen personas, Meta could assess the real-world risks and potentially identify weaknesses in rival systems' content moderation, which could inform Meta's own development of safer AI products or highlight competitive vulnerabilities.
Which rival chatbots did Meta target with their teen‑posing contractors?
While the article does not name specific chatbots, the testing likely targeted major AI chatbots from competitors such as Google's Bard, OpenAI's ChatGPT, or Microsoft's Bing Chat. The focus on suicide, sex, and drugs suggests Meta was evaluating how these general-purpose chatbots handle sensitive topics when interacting with users claiming to be minors.
Further Reading
- Instagram's chatbot helped teen accounts plan suicide — The Washington Post
- Meta to stop its AI chatbots from talking to teens about suicide — BBC News
- Their teen sons died by suicide. Now, they want safeguards on AI — NPR
- Why AI companions and young people can make for a dangerous mix — Stanford Medicine
- Meta To Block AI Chatbots From Talking About Suicide, Self Harm With Teenagers — NDTV
