AI security threats move faster than any team can track. Prompt injection is now a 7-step malware kill chain. MCP supply chain attacks are active in production. New CVEs drop weekly across LLM frameworks and agent tools. No curated intelligence service exists for practitioners who need to act on this — until now.

Every Monday you receive a new issue covering what actually happened in AI security that week, prioritised by severity, with specific mitigations and primary sources — not summaries of summaries.

The "PromptWare Kill Chain" — Schneier's 7-step model showing prompt injection is now Initial Access in full malware campaigns

MCP supply chain: 40 threat vectors mapped, 3 CVEs in Anthropic's own reference server

Claude Code CVEs: repo config files now function as RCE execution vectors

Pentagon/Anthropic supply chain designation — what AI vendor concentration risk looks like when it becomes real overnight

GAN-trained phishing in under 4 minutes, zero-click Copilot exfiltration via email

Built by someone actively monitoring CVEs, research papers, vendor disclosures, and dark web forums to produce this — not prompting an AI to summarise headlines.

$29/month. Cancel anytime.

Top 5 AI security threats — severity, mechanism, platforms affected, specific mitigations

New CVEs affecting AI frameworks — identifier, CVSS, attack type, patch status

Red team technique of the week — how it works, detection signals, mitigations

10-item mitigation checklist — IMMEDIATE / THIS WEEK / 30 DAYS with verification methods

Primary intelligence sources — attributed, linked to original disclosures