Skip to content
Source · Daily Brief

AI Daily Brief — 11 December 2025

Thursday compressed four landmark events into a single news cycle — among the most consequential days in AI history. OpenAI launched GPT-5.2 family (Instant + Thinking + Pro) across ChatGPT + API; August 2025 knowledge cutoff; speed/structured-reasoning/maximum-accuracy variants. GPT-5.2 Thinking 70.9% on GDPval (vs GPT-5.1 Thinking 38.8%) across 44 occupations — beats/ties top professionals; first model near 100% accuracy on 4-needle MRCR out to 256K tokens; SWE-Bench Pro 55.6% (Gemini 3 Pro 43.3%, Claude Opus 4.5 52.0%). API $1.75/$14 per Mtoken + cached input $0.18/M (~90% discount), 400K context + 128K max output. Altman to CNBC: 8-week “Code Red” sprint triggered by Gemini 3 should wind down by January. Disney announced $1B equity investment + warrants in OpenAI + 3-year Sora character licensing (exclusive year one) for Mickey + Pixar + Marvel + Star Wars characters + costumes + props + environments in early 2026. Talent likenesses/voices excluded; Disney IP cannot train models. Disney retains generated content ownership + can curate user clips onto Disney+. Iger to CNBC: “opportunity, not a threat” — “we’d rather participate than be disrupted.” Disney also becomes major OpenAI API customer + rolling ChatGPT to employees. Trump signs EO 14365 “Ensuring a National Policy Framework for AI” — directs federal agencies to challenge/deter state AI laws as fragmenting innovation. AI Litigation Task Force at DOJ to challenge “unconstitutional, preempted, or otherwise unlawful” state AI laws — particularly those compelling alterations to truthful outputs (First Amendment grounds). Within 90 days Commerce must issue Policy Notice making states with “onerous” AI laws ineligible for remaining BEAD “nondeployment” funds from $42B Broadband Equity Access and Deployment program. Google ships Gemini Deep Research agent + new Interactions API public beta in AI Studio — powered by Gemini 3 Pro; iteratively plans/searches/identifies knowledge gaps/synthesizes long reports with granular citations across web + document inputs. Google open-sources DeepSearchQA benchmark (900 tasks across 17 disciplines) — Gemini Deep Research 66.1% ahead of OpenAI Deep Research + Perplexity Deep Research; Kaggle leaderboard. ChatGPT Images relaunch with new flagship model — more precise edits preserving fine details, generation up to 4× faster. Harness $240M Series E at $5.5B ($200M primary + $40M tender, Goldman Sachs lead with IVP + Menlo + Unusual) for AI DevOps platform — on track to exceed $250M ARR in 2025. Meta plaintiffs file to add contributory infringement claim in AI-training copyright suit. TechCrunch frames Dec 11 as one-day OpenAI-vs-Google showdown ending the Nov 17 → Dec 11 cluster (Grok 4.1 + Gemini 3 + Opus 4.5 + GPT-5.2 inside 25 days).

Top stories

  • OpenAI launches GPT-5.2 (Instant + Thinking + Pro). Aug 2025 cutoff. 400K context + 128K output. Speed/structured-reasoning/max-accuracy variants. via OpenAI
  • GPT-5.2 Thinking 70.9% GDPval (vs 38.8% GPT-5.1) + first ~100% on 4-needle MRCR 256K + SWE-Bench Pro 55.6%. Beats Gemini 3 Pro (43.3%) + Opus 4.5 (52.0%) on SWE-Bench Pro. via OpenAI System Card
  • Altman: “Code Red” winds down by January. 8-week sprint triggered by Gemini 3’s LMArena + GPQA Diamond lead. “When a competitive threat happens, you want to focus on it, deal with it quickly.” via CNBC
  • Disney $1B equity + warrants + 3-year Sora licensing. Mickey + Pixar + Marvel + Star Wars characters + costumes + props + environments early 2026. Talent excluded; Disney IP cannot train models. Disney+ curates user clips. via OpenAI
  • Bloomberg: Disney $1B first major IP license for AI video. Equity + warrants for more equity. Disney becomes major API customer + ChatGPT for employees. via Bloomberg
  • Iger: “opportunity, not a threat”. “We’d rather participate than be disrupted.” Younger-audience reach via Sora social-style surface. via Fortune
  • Trump signs EO 14365 “Ensuring a National Policy Framework for AI”. Federal agencies challenge/deter state AI laws; uniform federal alternative. via White House
  • EO 14365 creates DOJ AI Litigation Task Force. Challenge “unconstitutional, preempted, or unlawful” state AI laws — particularly those compelling alterations to truthful outputs (First Amendment grounds). via Skadden
  • EO 14365 conditions BEAD broadband funding on state AI-law posture. Within 90 days Commerce issues Policy Notice; states with “onerous” AI laws ineligible for remaining “nondeployment” funds from $42B BEAD. via Sidley
  • Google ships Gemini Deep Research + Interactions API beta in AI Studio. Powered by Gemini 3 Pro; iteratively plans/searches/identifies gaps/synthesizes long reports with citations. via Google
  • Google open-sources DeepSearchQA (900 tasks / 17 disciplines). Gemini Deep Research 66.1% ahead of OpenAI + Perplexity Deep Research. Kaggle leaderboard. via Google DeepMind
  • GPT-5.2 API priced $1.75/$14 per Mtoken + cached $0.18/M (~90% discount). 400K context, 128K output. Cheaper input vs GPT-5, pricier output. via LLM-Stats
  • ChatGPT Images relaunch with new flagship model. More precise edits preserving fine details; generation up to 4× faster. via OpenAI
  • Harness $240M Series E at $5.5B (Goldman Sachs lead). $200M primary + $40M tender; IVP + Menlo + Unusual in. AI DevOps platform; on track >$250M ARR 2025. via TechCrunch
  • Meta plaintiffs add contributory infringement claim. Authors’ motion to amend complaint — keeps pressure on frontier fair-use defenses post-Bartz. via Copyright Alliance
  • TechCrunch: Dec 11 one-day OpenAI-vs-Google showdown. Caps Nov 17 → Dec 11 cluster: Grok 4.1 + Gemini 3 + Opus 4.5 + GPT-5.2 in 25 days = densest frontier-model concentration of 2025. via TechCrunch

Who shipped

OpenAI shipped GPT-5.2 + ChatGPT Images relaunch + Disney deal. Google shipped Deep Research + Interactions API + DeepSearchQA benchmark. Disney shipped $1B + 200+ character license. White House shipped EO 14365 + DOJ AI Litigation Task Force. Harness shipped $5.5B round.

Open-source pulse

DeepSearchQA’s open-leaderboard format on Kaggle structurally important — Google making research-agent benchmarking a public good. Anthropic/Mistral/Chinese labs quiet on launches.

Money, infra & hardware

Disney $1B + Harness $240M = $1.24B+ direct AI capital Thursday. Disney equity stake into OpenAI structurally unusual — first major studio with frontier-lab equity.

Quiet corners

EO 14365’s BEAD conditionality + Litigation Task Force are the substantive policy shock — first major Trump-era federal action preempting state AI laws; case-law battle expected through 2026.

By the numbers

  • 70.9% / 38.8% / 55.6% / ~100% MRCR 256K — GPT-5.2 Thinking GDPval / GPT-5.1 Thinking / SWE-Bench Pro / long-context recall
  • $1B / 200+ / 3 years — Disney OpenAI investment / characters / Sora licensing horizon
  • $42B / 90 days — BEAD broadband fund subject to conditionality / Commerce Policy Notice deadline
  • 66.1% / 900 / 17 — DeepSearchQA Gemini / tasks / disciplines
  • $240M / $5.5B — Harness round / post-money
  • Most-mentioned company: OpenAI
  • Quietest segment: Anthropic + Chinese labs launches

Compiled by AI Feed’s editor from verified web sources for 11 December 2025.