AI Daily Brief — 17 January 2025
Friday delivered a unanimous Supreme Court, a White House staffing pick, an FTC inquiry result, and a Google DeepMind paper that would influence the year’s inference-time scaling literature. The Supreme Court upheld the TikTok divest-or-ban law per curiam; the FTC’s 6(b) staff report on cloud-AI partnerships landed; the incoming Trump White House announced Sriram Krishnan as OSTP Senior Policy Advisor for AI; and Mind Evolution showed that an evolutionary inference-time search could solve more than 98% of TravelPlanner and Natural Plan instances with Gemini 1.5 Pro and no formal solver.
Top stories
- Supreme Court unanimously upholds TikTok divest-or-ban law. The per curiam opinion rejected First Amendment challenges and required ByteDance to divest by Jan 19 or face a US ban — a national-security finding rooted in Chinese-ownership concerns over data and recommendation algorithms. via SCOTUSblog
- FTC publishes staff report on AI cloud partnerships. The Office of Technology’s 6(b) study of partnerships between hyperscalers (Alphabet, Amazon, Microsoft) and frontier labs (Anthropic, OpenAI) flagged structural concerns: compute-access lock-in, engineering-talent capture, and switching costs that could harm AI-developer competition. via FGS Global
- Sriram Krishnan named White House Senior Policy Advisor for AI. The former a16z partner with product roles at Microsoft, Twitter, Meta, and Snap will sit at OSTP — the most consequential AI staffing pick announced ahead of Trump’s Monday inauguration. via FGS Global
- Google DeepMind ships “Evolving Deeper LLM Thinking” — Mind Evolution. An evolutionary inference-time search strategy that generates, recombines, and refines candidate responses; with Gemini 1.5 Pro it solves over 98% of TravelPlanner and Natural Plan instances without any formal solver. A foundational paper for the year’s inference-time scaling literature. via arXiv
- NYT v. OpenAI: magistrate grants nine sealing motions; RIAA v. Suno/Udio Special Master reports cooperative progress. Two parallel litigation tracks moving on procedural rails — the NYT case continuing its march toward substantive merits and the music-AI cases working through ESI, confidentiality, and source-code protocols. via McKool Smith tracker
Who shipped
Frontier-lab products: quiet. The week’s biggest names — OpenAI, Anthropic, Google DeepMind, Meta, xAI — sat the Friday out on the product side; the FTC report and the Supreme Court ruling were the day’s loud signals.
Open-source pulse
No new releases. DeepSeek was three days away from R1; Qwen was holding for late January.
Quiet corners
TSMC follow-up Friday commentary highlighted accelerating Blackwell ramp orders driving advanced-node utilisation. The Diffusion Rule continued reverberating: Anthropic‘s Dario Amodei joined former Trump deputy NSA Matt Pottinger in a WSJ op-ed publicly backing it — a notable split with NVIDIA, Oracle, and Microsoft’s opposition.
By the numbers
- Unanimous — SCOTUS vote on TikTok divest-or-ban
- 3 hyperscalers and 2 frontier labs probed by FTC’s 6(b) study
- 98%+ — Mind Evolution solve rate on TravelPlanner with Gemini 1.5 Pro
- Most-mentioned actor: Supreme Court
Compiled by AI Feed’s editor from verified web sources for 17 January 2025.