Skip to content

Updates log

This page aggregates content-level material updates. Code-level changes live in the Git history.

2026-04

Section titled “2026-04”
  • 2026-04-23 — Frontier-lab subpages filled out: Anthropic, OpenAI, Google DeepMind, ByteDance, and DeepSeek now each have substantive analysis across all five standalone subpages (Usage Policy, Model Card, Safety Framework, Transparency Report, Red-team and Evaluation Disclosures), approximately 1,000–1,300 lines per company. A unified top-bar component (SiteHeader) now runs across the homepage and all Starlight internal pages; search and the ⌘K shortcut are wired to pagefind.
  • 2026-04-22 — The corporate-practice axis expanded to 13 companies (added NVIDIA for the infrastructure layer; added a global dimension to the ByteDance–TikTok page; added a retrospective on DeepSeek’s “January 2025 moment”). In-depth analysis of Anthropic RSP v3 (2026-02) and interpretation of Google DeepMind FSF v3 (2026-04).
  • 2026-04-20 — Site launched. Initial topic coverage: risk classification, frontier GPAI, data and training, synthetic-content labelling (four topics × three jurisdictions). Top-level rule coverage: China’s five-tier hierarchy; the US federal layer (including the Trump 2025 series of executive orders); EU AI Act / GDPR / DSA primary-source index. Subnational coverage: Shenzhen; Colorado / California SB 53 / Texas TRAIGA / NYC Local Law 144; Spain AESIA / France CNIL.