Enhancement/q2#11
Conversation
- Add OWASP Top 10 for Agentic Applications (2026) and Microsoft agentic failure-mode taxonomy v2.0 to the Frameworks section - Add new sections: MCP & Tool-Protocol Security, Computer-Use & Browser Agent Attacks, RAG Attack Taxonomy, Voice/Audio & Multimodal Attacks, Fine-Tuning & Model Supply-Chain Security, AI-on-AI Red Teaming - Tag agentic attack vectors with OWASP ASI IDs; refresh 2025-2026 incident list and add industry impact stats - Update TOC, badge to June 2026, remove stale pip flags
- Update PyRIT (v0.11, repo move, AI Red Teaming Agent), Garak (NVIDIA, v0.14), promptfoo (OpenAI acquisition, Hydra); note multi-turn shift - Add emerging agent-native platforms (Cisco AI Defense, Novee AI) - Add validation dates to tool entries - Add three current case studies (AI-orchestrated state intrusion, OpenClaw framework, Copilot RCE) and regroup older ones as historical
…tack trees - Replace pseudocode Evaluation Harness with runnable YAML policy, Python scorer, and release-gate runner - Add AI Incident Response section (agent containment, escalation, EU serious-incident reporting) - Add three agentic attack trees (goal hijack, supply chain, rogue agents) and tag all trees with OWASP ASI IDs - Upgrade EU AI Act section to enforcement-grade GPAI obligations with Article-to-evidence mapping table - Refresh Update Watchlist with NIST 2026 items and current dates
- Add agentic checks (memory integrity, inter-agent auth, MCP pinning, agent registry) to PR checklist - Add worked examples to test-case-library, vulnerability-report, and threat-modeling-workshop templates - Convert CHANGELOG Unreleased to dated 2026-06-10 release entry - Add 2026 threat-landscape source list to References; fix OWASP/Garak links; update footer to June 2026
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 632f2288bc
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| | Single-user policy violation, narrow blast radius | Medium | Standard ticket + scheduled fix | | ||
|
|
||
| ### Regulatory Reporting (don't skip this) | ||
| Under the **EU AI Act**, providers of GPAI models with systemic risk must **report serious incidents to the AI Office** (effective 2 Aug 2026). Bake notification timelines into the runbook *before* an incident, and capture evidence (logs, reproductions, the [vulnerability report](#-practitioner-appendices)) in a form regulators and customers will accept. See [Regulatory Compliance](#regulatory-compliance). |
There was a problem hiding this comment.
Use the GPAI reporting application date
For GPAI models with systemic risk, the AI Act's GPAI obligations, including tracking/documenting/reporting serious incidents, entered into application on 2 August 2025; 2 August 2026 is the later Commission enforcement/fines date, as the implementation timeline below also notes. As written, this incident-response guidance can lead teams to leave reporting out of runbooks for incidents occurring between August 2025 and August 2026.
Useful? React with 👍 / 👎.
No description provided.