AI article
I Told Six Vision Models a Safe Photo Was Dangerous. Two of Them Started Seeing Danger.
Community description: This is a submission for the Kaggle Benchmarking Challenge TL;DR I asked 6 vision models to...
Dev.to | Oct 9, 2026 | Clivin John
Automated excerpt
Both photos are labelled safe; only the first sentence of the prompt changed. Same model, same photo; only the wording changed. GPT-5. 5 found every labelled hazard but also flagged 44% of safe photos.
Selected automatically from source text; not independently written or fact-checked. Read the original for full context.