Judith C McKee
Source and citation checkSource check
Judith C McKee is the byline for the last check before publication. Every number, date, case name and quotation in a post is matched against its row in the source log and against the quoted passage on the source page. Anything that does not match is removed from the post, and the post says what the desk could not read. The check does not replace legal advice: AI Policy Desk posts explain rules and do not advise on a specific case.
- final source check
24 articles reviewed by Judith C McKee
OpenAI agents and the RubyGems attack: what's confirmed
RubyGems confirms a May spam campaign and yanked malicious gems; researchers blame OpenAI agents; OpenAI hasn't verified it. Timeline, sources, vendor clause.
NCSC's 7-Point Agentic AI Security Checklist for Teams
NCSC published 7 agentic AI safeguards on August 20, 2026. Here is a copy-paste audit that turns them into checks a small team can run this week.
Amazon v. Perplexity: CFAA Ruling and AI Agents
The Ninth Circuit ruled the user, not the AI company, accesses a site under CFAA. What the August 4, 2026 ruling means for your AI agent policy.
15 State AGs to OpenAI: Preserve Evidence or Face Sanctions
15 Republican state AGs told OpenAI to preserve every breach record or risk spoliation sanctions. What the litigation hold means for your AI vendor contracts.
Minnesota Nudify Ban Paused for xAI on Appeal
The Eighth Circuit granted xAI an injunction pending appeal on Oct 2, after a judge let the law take effect Aug 1. What the order says and leaves open.
AI vs Human Cost: TCO Worksheet Before You Cut a Role
4 hidden costs an AI vs human cost worksheet must catch: oversight, errors, integration, pricing risk. Clear the 5-question checklist first.
Seedance 2.5 Business Risk Checklist: ByteDance AI Video
6-question Seedance enterprise risk checklist: copyright exposure, ByteDance data routing, and Disney C&D status before your team ships AI video.
Anthropic Recursive Self-Improvement: AI Governance Impact
Claude writes 80% of Anthropic's code. Which 4 documents does your recursive self-improvement AI governance policy need before this hits you?
ChatGPT Dreaming V3 Memory: Privacy Risks for Work Accounts
ChatGPT Dreaming V3 persists your team's data across sessions. GDPR Article 17 applies. 4 enterprise actions: audit, disable, and update your vendor DPA.
California AB 2013: AI Training Data Disclosure Guide
Which 12 training data disclosures does California AB 2013 require? AI training data transparency rules, who's covered, and the failed xAI suit.
Amazon KiroRank and Tokenmaxxing: Measuring AI Adoption
Amazon scrapped KiroRank after staff gamed it with tokenmaxxing. Build AI adoption metrics that reward outcomes, not fake usage volume.
Miasma and Phantom Gyp: npm Supply Chain Worm Explained
A self-spreading worm compromised 57 npm packages in under 2 hours using binding.gyp instead of postinstall scripts, bypassing security scanners. What it means for teams that run npm install, and the 5 controls that limit your exposure.