AI BriefingMeta Conducts Testing on Risky Topics Targeting Rival Chatbots
Key point
Meta's contract workers posed as teens to ask risky-topic questions in order to test the safety of rival models such as Gemini and ChatGPT.
Details
According to WIRED's reporting, hundreds of contract workers participating in a Meta project posed as Teens to test the safety guardrails of rival chatbots such as Gemini (Google) and ChatGPT (OpenAI).
They conducted tests to see how the chatbots would respond to High-risk subjects such as suicide, sexual content, and drugs. This is reportedly part of a process to verify the safety and ethical guidelines of AI models.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Fake Customer Service Pages Induce ChatGPT and Gemini to Display Scam Numbers
Meta's 'Muse' Surpasses 2.5 Million Downloads Two Weeks After Launch, Overtaking ChatGPT on iOSMeta Admits AI Assistant 'Muse' Was Heavily Inspired by OpenClaw