Researchers introduced Latent Frequency Masking, an attack that erases AI-generated image watermarks by manipulating the image's latent representation. The attack preserves image quality and is more efficient than existing methods. This highlights the need for robust watermarking methods and includes latent-frequency manipulation in security evaluations.

A study finds that safety routing in LLMs can be broken by distribution shift and that the current evaluation methods may not accurately reflect the true safety of the models. The study suggests that safety routing should be evaluated under shift and against a baseline chosen without test labels, and that recognition-based defenses should be scored on harm against an attacker who chooses what the model sees.

Researchers investigated how weak reviewers can audit strong coding agents. They found that official execution evidence can improve defect catch and reduce over-rejection. The study used 411 execution-labeled traces from three agents and 101 controlled cases.

A study evaluates the environment specification capabilities of large language models (LLMs) in generating software code. The research highlights systematic generalization failures in current LLMs, leading to inconsistent, redundant, or incomplete dependency specifications. This affects the portability and execution of generated code.

Researchers presented a deterministic AI security risk assessment framework that uses a Control ID taxonomy and formal verification to evaluate AI systems. The framework normalizes heterogeneous artifacts into a project-independent taxonomy and outputs technique-indexed feasibility and impact levels. It was evaluated on five public open-source AI projects, showing consistent downward shifts in feasibility profiles under strengthened observable controls.

This paper introduces ShieldCLIP, a framework for selective safety alignment in multimodal foundation models like CLIP. It conditions safety alignment on the observed safety state of each modality, preserving safe content and redirecting only unsafe content. The authors also introduce ViSUv2, a 195k-quadruplet dataset with independent per-modality safety labels. ShieldCLIP achieves consistent reductions in harmful outputs in various tasks and settings.

This paper presents a new approach to embedding security properties into AI-enabled Cyber-Physical Systems (CPS) by integrating Signal Temporal Logic (STL) specifications into forecasting models. This allows predictive models to enforce system-level constraints during inference and mitigate adversarial perturbations.

Cloudflare introduces a framework to address emerging AI-driven security threats, emphasizing the need for overlapping controls, continuous validation, and faster mechanisms to correlate and contain suspicious behavior.

A paper compares the performance of a purpose-built security context layer, Sola Security Brain, with a general-purpose coding agent, Claude Code, in cloud-security investigation tasks. Sola Security Brain shows a 79.2% relative gain in coverage over Claude Code, with lower reasoning cost per task.