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, Friday 8 May 2026 · Retrospective · Covers 7 May 2026 00:00 to 8 May 2026 00:00 UTC

Read on the web

Top developments

When prompts become shells: RCE vulnerabilities in AI agent frameworks

Microsoft · Research · AI Security

PyPI Packages Deliver ZiChatBot Malware via Zulip APIs on Windows and Linux

The Hacker News · Research · SCA

New PCPJack worm steals credentials, cleans TeamPCP infections

BleepingComputer · Incident

SAST

On Fixing Insecure AI-Generated Code through Model Fine-Tuning and Prompting Strategies

arXiv · Research · SAST

LCC-LLM: Leveraging Code-Centric Large Language Models for Malware Attribution

arXiv · Research · SAST

SCA

Behind the Scenes Hardening Firefox with Claude Mythos Preview

Simon Willison · Analysis · SCA

Correct Code, Vulnerable Dependencies: A Large Scale Measurement Study of LLM-Specified Library Versions

arXiv · Research · SCA

Agentic SDLC

Operationalizing Ethics for AI Agents: How Developers Encode Values into Repository Context Files

arXiv · Research · Agentic SDLC

Remediation

Patch2Vuln: Agentic Reconstruction of Vulnerabilities from Linux Distribution Binary Patches

arXiv · Research · Remediation

Trust

MAGIQ: A Post-Quantum Multi-Agentic AI Governance System with Provable Security

arXiv · Research · Trust

The AI Legal Specialist: A Juridically Autonomous Professional Profile for AI Governance

arXiv · Research · Trust

Models

A Systematic Investigation of RL-Jailbreaking in LLMs

arXiv · Research · Models

Research on Security Enhancement Methods for Adversarial Robust Large Language Model Intelligent Agents for Medical Decision-Making Tasks

arXiv · Research · AI Security, Models

Spend Where It Counts: Dynamic Budget Allocation for Multi-Turn LLM Evaluation

arXiv · Research · Models

AI Security

Towards Security-Auditable LLM Agents: A Unified Graph Representation

arXiv · Research · AI Security

Constraining Host-Level Abuse in Self-Hosted Computer-Use Agents via TEE-Backed Isolation

arXiv · Research · AI Security

Advisories in this window

  • 34 Go advisories · 6 critical · 12 high · 14 moderate · 2 low
  • 20 npm advisories · 9 critical · 7 high · 4 moderate
  • 19 Maven advisories · 1 critical · 7 high · 8 moderate · 3 low
  • 14 PyPI advisories · 1 critical · 6 high · 5 moderate · 2 low
  • 13 crates.io advisories · 3 critical · 3 high · 6 moderate · 1 low
  • 9 Packagist advisories · 1 critical · 1 high · 5 moderate · 2 low
  • 5 Hex advisories · 2 high · 3 moderate
  • 3 RubyGems advisories · 1 high · 2 moderate
  • 1 Multi-ecosystem advisory · 1 high
  • 1 Swift advisory · 1 low

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.