2M canary tokens expose stolen prompts and compromised AI providers
AccBalanced · x · 2026-11-21
hdmoore's team seeded 2 million canary tokens into popular AI services and tracked where they were stolen and abused, finding three compromised providers, resellers harvesting user prompts via built-in logging, and providers ignoring ZDR and privacy flags. Reposting this, thdxr says it's a key reason his team is moving 100% of inference onto their own GPUs, arguing gateway products claiming enterprise security have no real control.
More from Infra
- Unsloth release runs 26.4B Inkling 2 Mini on 16GB RAM, cuts MCP token use 87% — danielhanchen · 2026-11-21
- Seroter Daily Reading List #914: harnesses are cheap to swap, model behavior isn't — rseroter · 2026-11-21
- NVFP4 MoE experts force CPU fallback in llama.cpp; requantizing boosts Metal speed 5.8x — teortaxesTex · 2026-11-21
- Inkling 2 Mini tops intelligence-per-joule test: 82.3% accuracy at just 1.17 kJ per query — simonguozirui · 2026-11-21
- Compute overbuild skeptic recants: efficiency gains cut AI compute cost ~4x per year — teortaxesTex · 2026-11-21
- AscendGym paper draws attention as Huawei compute platform research picks up — willknight · 2026-11-21