AI Security Threat Feed

PyTorch: Dropout inconsistency enables membership inference

PyTorch: OOB write causes incorrect bitwise shift results

PyTorch: torch.compile silent output inconsistency

PyTorch: reachable assertion in nn.Fold with inductor

PyTorch: PairwiseDistance silent miscalculation, integrity risk

n8n: stored XSS in LangChain chat trigger (public)

Transformers: ReDoS in EnglishNormalizer exhausts CPU

xgrammar: DoS via oversized JSON schema grammar parsing

picklescan: scanner bypass enables model RCE

picklescan: scanner bypass enables RCE via ML model files

picklescan: scan bypass allows silent RCE via ML models

picklescan: detection bypass allows undetected RCE in ML models

picklescan: scanner bypass enables pickle RCE in ML models

picklescan: scanner bypass allows pickle-based RCE

picklescan: scanner bypass enables RCE via ML models

picklescan: detection bypass allows pickle RCE in ML pipelines

picklescan: scanner bypass leads to undetected RCE

picklescan: scanner bypass enables PyTorch gadget RCE

picklescan: scanner bypass enables RCE via model files

picklescan: RCE bypass enables ML supply chain attack