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 12 July 2026 · Retrospective · Covers 11 July 2026 00:00 to 12 July 2026 00:00 UTC

Read on the web

Top developments

'Ghostcommit' hides prompt injection in images to fool AI agents, steal secrets

BleepingComputer · Research · Agentic SDLC, AI Security

Compromised jscrambler 8.14.0 npm Release Drops Rust Infostealer During Install

The Hacker News · Incident · SCA

Australia warns of global campaign targeting vulnerable CMS platforms

BleepingComputer · Incident

SCA

Which Neurons Detect Malicious Code? A Probing Study of LLM Security Knowledge

arXiv · Research · SCA

Trust

Falsifiable Release Gates for Self-Improving Systems: Standing Invariants at Scale

arXiv · Research · Trust

Models

When Are Sparse Feature Interventions Actually Localized? Matched Evaluation for SAE-Based Safety Control

arXiv · Research · Models

Minionese: Comprehensive Benchmark and Mechanistic Study of Multilingual LLM Safety

arXiv · Research · Models

AI Security

NetInjectBench: Benchmarking Indirect Prompt Injection in Tool-Using Large Language Model Agents for Network Operations

arXiv · Research · AI Security

Devil in the Lens: Analyzing and Defending Physical Prompt Injection Against Vision-Language Models on Wearable Devices

arXiv · Research · AI Security

Advisories in this window

  • 2 npm advisories · 2 critical
  • 1 Maven 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.