Skip to content
Source · Daily Brief

AI Daily Brief — 27 December 2025

Saturday’s only structural news was China policy. Cyberspace Administration of China (CAC) released draft “Interim Measures for the Management of AI Human-like Interactive Services” covering chatbots/companions that simulate personality. Requirements: AI-identity disclosure at login + every 2 hours; addiction safeguards; extra protections for minors + elderly; full-lifecycle provider responsibility; limits on emotional/sensitive data for training. Public comment open until Jan 25. ChinaLawTranslate published full English text — first detailed look at scope + obligations for Western readers. Otherwise quiet day — no frontier-lab releases, no funding rounds, no notable open-source drops from US/EU labs. Most Western coverage was year-end recap content (PBS NewsHour “Year in AI” segments + Yahoo’s “Google ended 2025 on top” retrospective).

Top stories

  • China CAC draft “Interim Measures for AI Human-like Interactive Services”. AI-identity disclosure at login + every 2 hours. Addiction safeguards. Minors + elderly extra protections. Full-lifecycle provider responsibility. Emotional/sensitive training data limits. Comment to Jan 25. via Bloomberg
  • Full CAC draft text published in English translation. First detailed look at scope + obligations for Western readers. via ChinaLawTranslate
  • Saturday otherwise quiet — only Western news year-end retrospectives. No frontier-lab releases. No funding rounds. No notable open-source drops US/EU. PBS NewsHour “Year in AI” + Yahoo “Google ended 2025 on top.” via Yahoo Finance

Who shipped

Nobody. OpenAI, Anthropic, Google, Meta, xAI, DeepSeek, Alibaba all dark. CAC’s draft = the only structural Saturday item.

Open-source pulse

No new drops. Friday’s GLM-4.7 + Qwen-Image-Layered + Fun-Audio-Chat continued community uptake.

Money, infra & hardware

No new rounds. NVIDIA-Groq $20B from Dec 24 still anchoring infra coverage.

Quiet corners

China CAC rule = substantive policy moment for AI companions — first major regulator framework targeting personality-simulating chatbots specifically.

By the numbers

  • 0 — frontier-lab launches Saturday
  • Every 2 hours / login — CAC AI-identity disclosure frequency
  • Jan 25 — public comment deadline
  • Most-mentioned topic: China AI companion regulation + year-end retrospectives
  • Quietest segment: everything except CAC

Compiled by AI Feed’s editor from verified web sources for 27 December 2025.