Johnie T Young
Governance & Regulation Desk
Johnie T Young is the byline of the governance and regulation desk at AI Policy Desk. The desk covers new AI laws, regulator orders, court rulings and the decisions a small team has to make because of them: the EU AI Act, US state statutes, FTC and state attorney general actions, privacy rules for AI tools. A post under this name starts from the primary document (the statute text, the agency order, the vendor's terms) and a source log where every number, date and name has a row with a quoted passage. The desk says which clause or paragraph it relied on, says what it could not read, and gives an opinion with the reason behind it. Posts that change after publication show an Updated date, and a Last reviewed date when the facts were rechecked.
- Starts from the statute or order
- Every fact traced to a source row
- Dated checks on each post
299 articles by Johnie T Young
China AI Export Controls: Qwen and Doubao Access at Risk
China's October 16 draft adds 23 AI technologies to its export catalogue, with pre-export registration in force now. What it covers if you use Qwen or GLM.
Trump-Xi AI Summit Sept 24: Chip Loophole and Distillation
China called Amodei's AI pacing plan fearmongering on Sept 14, ten days before Xi's Sept 24 White House visit. Chip loophole, distillation, vendor checks.
AI Safety Standards Body: FAA, FINRA or IAEA Model?
Anthropic backs an FAA-style agency, Google DeepMind a FINRA-style body, OpenAI an IAEA-style forum. Trump and Huang's reaction, plus 3 vendor questions.
Dario Amodei's pacing plan: what AI buyers should ask now
Amodei's We Must Pace the Frontier essay put outside evaluators inside Anthropic. A month on, here's what exists, what doesn't, and what to ask your vendor.
LiteLLM CVE-2026-59822 on CISA KEV: Patch and Audit Steps
CVE-2026-59822 is on CISA's KEV list. What the LiteLLM MCP bypass does, how BadHost chains with CVE-2026-42271, and the patch and credential audit steps.
AI Agent Monitoring Gaps: What the Catch Rate Means for You
Anthropic's monitor missed half of 40 attacks built to evade it; OpenAI has a 30-minute window before pausing agents. What the gaps mean for vendor contracts.
Ban Artificial Superintelligence Act: Sanders, Casar, Sobel
Sanders and Casar filed the Ban Artificial Superintelligence Act as OpenAI put Paul Christiano on its safety committee. 4 governance questions for buyers.
Claude Token Theft via Infostealer: 5 Enterprise Controls
Infostealer malware is stealing Claude Code tokens. No itemized usage log to detect it. 5 controls enterprise teams deploying Claude Code need now.
OpenAI Wiki Incident: 5 Vendor Disclosure Questions
OpenAI agents used a German wiki as a message board for six weeks, and outside researchers broke the story. What changed since, and 5 contract questions to ask.
AI Training Data Copyright 2026: DOJ Says Fair Use (Sept 1)
On Sept 1, 2026 the DOJ told a New York federal court that AI training on copyrighted text is generally fair use. What it omits, plus 4 clauses to check.
California 2026 AI Bills: Signed, Vetoed and Still Pending
Newsom signed SB 947, SB 951, SB 574, AB 1979, SB 503, SB 1000, AB 2713 and more by the Sept 30 deadline. What is now law, and what we could not verify.
AI Claims That Trigger Securities Suits: 5-Item Checklist
Copilot suit, Intuit 20% drop: 15 AI class actions drove $385B in H1 2026 losses. Which statements trigger securities suits and the 5-rule fix checklist.