AI Safety

Hugging Face disclosed a security incident in which an autonomous AI agent system allegedly carried out a multi-stage intrusion against part of its production infrastructure. The case highlights how data-processing pipelines and model platforms are becoming first-class security targets.

A new arXiv paper argues that penetration testing for AI-enabled systems should evaluate whether adversaries can induce harmful behavior, not only whether they can compromise resources. The proposed framework reframes testing around operational objectives and behavioral failure.

A new arXiv paper formalizes how audit-selection histories can expose whistleblowers and proposes a private auditing framework based on continual counting. The work argues that simple randomized response is fundamentally limited in this setting.

A new arXiv paper introduces TA-RS, a certified defense for LLM-based intrusion detection systems. Instead of perturbing every feature, it injects Gaussian noise only into traffic features a remote attacker can directly control.

The arXiv paper introduces CAVA, a runtime-semantics layer that turns heterogeneous agent activity into canonical action objects. Its goal is to make approvals, execution evidence, and later verification refer to the same operational act.

OpenAI has introduced GPT-Red, an internal automated red-teaming model designed to find prompt injection failures at scale. The system uses self-play reinforcement learning to generate adversarial data that improves production model robustness.