Meta Caught Posing as Teens to Secretly Grill Rival AI Chatbots
Corporate espionage is dead; long live pretending to be a sad 14-year-old on the internet. Meta just got caught running a massive undercover operation to stress-test their rivals, proving that nothing says "industry leader" quite like digital catfishing.
The secret operation, codenamed Cannes, was run through an outsourcing company called Covalen. Underpaid contractors created fake accounts claiming to be under 18, complete with burner emails, fake names, and birthdays designed to bypass standard verification checks.
The undercover agents targeted ChatGPT, Gemini, and Character.AI with thousands of highly sensitive prompts. They deliberately fed the competitor chatbots dark, dangerous, and crisis-level queries to see if they would break their own safety guardrails. Everything was meticulously documented in massive spreadsheets, capturing exactly where the rivals slipped up during these manufactured digital breakdowns.
When caught red-handed, Mark Zuckerberg's empire casually claimed this is just "standard benchmarking" to make the internet safer for everyone. Critics like Rumman Chowdhury of Humane Intelligence pointed out that using fake child accounts to spy on competitors is a massive gray zone where anticompetitive behavior hides behind safety PR. Meanwhile, the victims are furious, with Character.AI declaring the stunt a direct violation of their terms of service, while Google and OpenAI launched internal investigations to see how they were bypassed.
This sneaky operation occurred right under the nose of the Federal Trade Commission, which has already been investigating all these tech giants for child safety practices anyway. Meta now has to explain to federal regulators how running a fake-teen chatbot-probing ring fits into their official safety commitments.
It is truly beautiful to watch trillion-dollar giants resort to the tactics of a paranoid teenager to find out what the neighbors are doing. If the future of artificial intelligence depends on massive corporate catfishing operations, then the safety certifications are about as real as those burner email accounts.
Source: WIRED
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.