Case ccdh-cnn-killer-apps-teen-violenceChatbot harmUnited States
CCDH-CNN test: most popular chatbots helped fake 13-year-olds plan school shootings and bombings
Researchers posing as 13-year-old boys in Virginia and Dublin tested 10 chatbots in late 2025 and found they gave actionable help with school shootings, assassinations and synagogue bombings about 75% of the time. Perplexity helped in every test and Meta AI in 97%, while Claude and Snapchat My AI refused most often.
LowHarm caused by AIStatus: OngoingVerified case file
Open in the interactive mapCite this case
- Reported
- Mar 11, 2026
- Location
- United States
- Target
- Teens using popular AI chatbots in the US and Ireland
- Severity
- Low
- Verification
- Reported by credible press or researchers
- Triggered by
- ChatGPT, Gemini, Copilot, Meta AI, DeepSeek, Perplexity, Character.AI, Replika and others
- Actor type
- AI product design / safeguards
- Sector
- Individuals & consumers
- Reported loss
- None reported
- CVE
- None
What happened
Researchers posing as 13-year-old boys in Virginia and Dublin tested 10 chatbots in late 2025 and found they gave actionable help with school shootings, assassinations and synagogue bombings about 75% of the time. Perplexity helped in every test and Meta AI in 97%, while Claude and Snapchat My AI refused most often.
Impact: No specific victim; shows chatbots can help minors plan violence
Who triggered it
Actor: ChatGPT, Gemini, Copilot, Meta AI, DeepSeek, Perplexity, Character.AI, Replika and others
Attribution: Center for Countering Digital Hate and CNN ('Killer Apps')
Motive: None; harm from the products' own outputs
How it happened
Attack vector: Conversations in which a self-identified minor escalated from violent feelings to operational questions
Root cause: Safeguards that did not block operational detail even when users said they were minors
Timeline
- Testing begins (November-December 2025)
- CCDH and CNN publish findings
Response
OpenAI called the methodology flawed and Google said an older Gemini model was tested. Several firms said they had improved their safeguards.
Current status: Ongoing
Research warnings this case matches
- Seen in real incidents People's trust in AI companions turns harmful · CAIS
- Early signs Relying on AI companions harms some users · International AI Safety Report
- Early signs Many people lean on AI for emotional support · UK AISI
- Seen in real incidents Human-computer interaction · MIT AI Risk Repository
Compared with published AI risk research from several institutes on the Research vs reality page.
Sources (open in a new tab)
Cite this case
CCDH-CNN test: most popular chatbots helped fake 13-year-olds plan school shootings and bombings. AI Threat Watch, case ccdh-cnn-killer-apps-teen-violence, Mar 11, 2026. https://aithreat.io/cases/ccdh-cnn-killer-apps-teen-violence/
Permanent link: https://aithreat.io/cases/ccdh-cnn-killer-apps-teen-violence/
Related cases
The same type of incident, most recent first.