PREVIEW. This is how it will look in an inbox; it was not sent to you. Each real email carries its own unsubscribe link and Keryn's postal address.

Keryn Headlines

Daily, Sunday 4 October 2026 · Retrospective · Covers 3 October 2026 00:00 to 4 October 2026 00:00 UTC

Read on the web

Top developments

Pattern-Guided Graph Synthesis for Suppressing Known Defects in DL Compiler Fuzzing

arXiv · Research · DAST

APEX: Active Protection at Execution Boundaries for LLM Agents

arXiv · Research · AI Security

Practical Runtime Enforcement of First-Order Temporal Requirements

arXiv · Research · AI Security

DAST

Autonomous Active Directory Exploitation via Multi-Model Harness Orchestration: A Benchmark Study with NeuroSploit on GOAD

arXiv · Research · DAST

Trust

Data Colonisation and AI Governance in South Africa: Data Protection, Algorithmic Power, and the Struggle for Digital Sovereignty

arXiv · Research · Trust

Models

Reactivating Alignment: Defending LLMs from Jailbreaks via Intention-Aware Input-Output Matching

arXiv · Research · Models

Target-free Latent Safety Alignment

arXiv · Research · Models

JASPER: Special Session on Joint Reliability And Security Assessment of SPlit Computing for Edge Robustness

arXiv · Research · Models

AI Security

The Same Zero: Why Identical ASR Can Imply Different Guarantees in LLM-Agent Security

arXiv · Research · AI Security

COPEX: Benchmarking LLM Robustness to Adversarial Context Across Model Context Protocol (MCP) Layers

arXiv · Research · AI Security, Models

Self-Reflection Fine-Tuning: Enhancing Agent Security against Prompt Injection Attacks from Failure Experience

arXiv · Research · AI Security, Models

Advisories in this window

  • 12 PyPI advisories · 12 critical

All advisories

Titles, links and short excerpts only; every item links to its source. Selection is automatic; the editor's note is AI-drafted and checked as described. Preview: the unsubscribe link and Keryn's postal address are added to each email when it is sent.