AI article
An AI agent guard claims zero successful attacks. I re-ran its tests.
Community description: The guard is OpenAPPA. I tested it on four AI models: what its zero counts, what it misses, and what...
Dev.to | Oct 12, 2026 | Andrey Gubanov
Automated excerpt
In the outside test, planted tricks often fooled agents without a guard. Without the guard, the agents got 92–97% of their jobs done. Without the guard, that happened once in 440 runs.
Selected automatically from source text; not independently written or fact-checked. Read the original for full context.