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 19 February 2026 · Retrospective · Covers 18 February 2026 00:00 to 19 February 2026 00:00 UTC

Read on the web

Top developments

CVE-2021-22175: GitLab Server-Side Request Forgery (SSRF) Vulnerability added to CISA's Known Exploited Vulnerabilities catalog

CISA · Advisory

Would You Click ‘Accept’? Automatically detecting malicious Azure OAuth applications using LLMs

Wiz · Research

Exact Certification of Data-Poisoning Attacks Using Mixed-Integer Programming

arXiv · Research · Models

SCA

An Evaluation of Large Language Models for Detection of Malicious Python Packages

arXiv · Research · SCA

Agentic SDLC

Can Adversarial Code Comments Fool AI Security Reviewers -- Large-Scale Empirical Study of Comment-Based Attacks and Defenses Against LLM Code Analysis

arXiv · Research · Agentic SDLC, Models

Remediation

AI Found Twelve New Vulnerabilities in OpenSSL

Schneier on Security · Analysis · Remediation

Models

Mind the GAP: Text Safety Does Not Transfer to Tool-Call Safety in LLM Agents

arXiv · Research · Models

DeepContext: Stateful Real-Time Detection of Multi-Turn Adversarial Intent Drift in LLMs

arXiv · Research · Models

NeST: Neuron Selective Tuning for LLM Safety

arXiv · Research · Models

AI Security

AgentLAB: Benchmarking LLM Agents against Long-Horizon Attacks

arXiv · Research · AI Security

Formal Policy Enforcement for Agentic Systems

arXiv · Research · AI Security

The Vulnerability of LLM Rankers to Prompt Injection Attacks

arXiv · Research · AI Security, Models

Advisories in this window

  • 42 npm advisories · 1 critical · 26 high · 15 moderate
  • 10 PyPI advisories · 1 critical · 5 high · 3 moderate · 1 low
  • 7 Go advisories · 2 high · 2 moderate · 3 low
  • 7 Packagist advisories · 2 high · 5 moderate
  • 3 Maven advisories · 1 high · 1 moderate · 1 low
  • 1 GitHub Actions 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.