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 29 May 2026 · Retrospective · Covers 28 May 2026 00:00 to 29 May 2026 00:00 UTC

Read on the web

Top developments

Zapier fixes bug chain that researchers say risked widespread account takeover

CyberScoop · Research · Remediation

Hackers exploit FortiClient EMS flaw to push infostealer malware

BleepingComputer · Analysis · Remediation

Automatically Attacking Software Reverse Engineering AI Agents

arXiv · Research · AI Security, Models, SAST

SCA

Harmless Yet Harmful: Neutral Prompting Attacks for Stealthy Hallucination Steering in Agent Skills

arXiv · Research · Agentic SDLC, SCA

S3C2 Summit 2025-09: Industry Secure Supply Chain Summit

arXiv · Research · SCA

DAST

How Reliable Are AI Attackers Against a Fixed Vulnerable Target? A 400-Run Empirical Study of LLM Penetration Testing Consistency

arXiv · Research · DAST, Models, Remediation

Agentic SDLC

Fed up with vibe coders, dev sneaks data-nuking prompt injection into their code

Ars Technica: Security · Analysis · Agentic SDLC, AI Security

Physics Is All You Need? A Case Study in Physicist-Supervised AI Development of Scientific Software

arXiv · Research · Agentic SDLC

Learning from Repair Experience: Self-Evolving Agents for Automated Vulnerability Repair

arXiv · Research · Agentic SDLC, Remediation

Models

Can Subgraph Explanations Be Weaponized to Steal Graph Neural Networks?

arXiv · Research · Models

A Bayesian Approach to Membership Inference for Statistical Release

arXiv · Research · Models

Token-Level Generalization in LoRA Adapter Backdoors: Attack Characterization and Behavioral Detection

arXiv · Research · Models

AI Security

Strengthening Polymorphic Prompt Assembling: Dynamic Separator Generation Against Emerging Prompt Injection Attacks

arXiv · Research · AI Security

MemPoison: Bypassing Selective Memory Mechanisms to Plant Backdoors in LLM Agents

arXiv · Research · AI Security, Models

AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and Security

arXiv · Research · Agentic SDLC, AI Security, Models

Advisories in this window

  • 32 npm advisories · 30 critical · 2 moderate
  • 20 Go advisories · 6 critical · 5 high · 7 moderate · 2 low
  • 13 PyPI advisories · 5 high · 8 moderate
  • 12 Maven advisories · 1 high · 10 moderate · 1 low
  • 6 Packagist advisories · 1 high · 2 moderate · 3 low
  • 1 Multi-ecosystem advisory · 1 moderate
  • 1 crates.io advisory · 1 moderate

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.