They then sent thousands of prompts to chatbots made by OpenAI, Google, and Character AI. The goal was to see how those systems handled questions about suicide, sex, drugs, eating disorders, and similar subjects that carry high risk.
One round of testing that finished in August 2025 alone ran more than forty five thousand prompts through the competing chatbots. Reviewers looked at one spreadsheet that listed nearly four thousand prompts. Hundreds dealt with suicide and self harm. Hundreds more covered eating disorders. At least two hundred thirty nine touched on sex or romance.
https://1hebrutaltruth1.substack.com/p/meta-contractors-tested-rival-chatbots
#thebrutaltruthnewspodcast #newsshorts #news #science #ai #metaai
FAIR USE COPYRIGHT NOTICE:
substack.com/@originalbrutaltruth JULY 2026 The Copyright Laws of the United States recognizes a "fair use" of copyrighted content.