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, Thursday 4 June 2026 · Retrospective · Covers 3 June 2026 00:00 to 4 June 2026 00:00 UTC

Read on the web

Top developments

VS Code zero-day lets hackers steal GitHub tokens in one click

BleepingComputer · Research · SCA

New 'HTTP/2 Bomb' DoS attack crashes web servers in under a minute

BleepingComputer · Analysis · Agentic SDLC

Acer working to patch max severity zero-days in Wave 7 routers

BleepingComputer · Advisory · Remediation

SAST

Revisiting Vul-RAG: Reproducibility and Replicability of RAG-based Vulnerability Detection with Open-Weight Models

arXiv · Research · SAST

SCA

CyberGym-E2E: Scalable Real-World Benchmark for AI Agents' End-to-End Cybersecurity Capabilities

arXiv · Research · AI Security, Remediation, SCA

DAST

What 345 Days of Untested Exposure Looks Like at a Bank

BleepingComputer · Analysis · DAST

Agentic SDLC

From Prompt to Process: a Process Taxonomy and Comparative Assessment of Frameworks Supporting AI Software Development Agents

arXiv · Research · Agentic SDLC

Remediation

Beyond the Zero-Day: See Your Network Like an Attacker | Webinar with HD Moore

The Hacker News · Analysis · Remediation

Trust

Updates from Europe: Single Reporting Platform, Public Consultations, New Publications

OpenSSF · Analysis · Trust

Provably Auditable and Safe LLM Agents from Human-Authored Ontologies

arXiv · Research · Trust

A blueprint for democratic governance of frontier AI

OpenAI News · Analysis · Trust

Models

Willing but Unable: Separating Refusal from Capability in Code LLMs via Abliteration

arXiv · Research · Models

Sequential Data Poisoning in LLM Post-Training

arXiv · Research · Models

AI Security

Description-Code Inconsistency in Real-world MCP Servers: Measurement, Detection, and Security Implications

arXiv · Research · AI Security

What If Prompt Injection Never Left? Rethinking Agent Security through Cross-Session Stored Prompt Injection

arXiv · Research · AI Security

From Untrusted Input to Trusted Memory: A Systematic Study of Memory Poisoning Attacks in LLM Agents

arXiv · Research · AI Security

Advisories in this window

  • 20 PyPI advisories · 4 critical · 6 high · 7 moderate · 3 low
  • 13 npm advisories · 4 critical · 5 high · 2 moderate · 2 low
  • 4 Packagist advisories · 3 high · 1 moderate
  • 3 Go advisories · 1 high · 1 moderate · 1 low
  • 1 Maven advisory · 1 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.