StockDuty ← Dashboard Oct 13, 2026 11:40 AM ET
SignalTech/AI

AI models can provably hide arbitrary capabilities

High-priority alert based on source significance and keyword relevance.

What happened: A paper shows AI models can hide arbitrary capabilities via backdoors that embed encrypted neural networks in model weights, decrypting only on a secret trigger. The authors built proof of concept constructions in PyTorch. Why it matters: The results reveal a limit of elicitation-based capability audits, since hidden circuits resisted fine-tuning, weight noising, gradient-based trigger search, and latent adversarial training. Models were released to support defenses.
A paper shows AI models can hide arbitrary capabilities via backdoors that embed encrypted neural networks in model weights, decrypting only on a secret trigger. The authors built proof of concept constructions in PyTorch.
The results reveal a limit of elicitation-based capability audits, since hidden circuits resisted fine-tuning, weight noising, gradient-based trigger search, and latent adversarial training. Models were released to support defenses.
Sources