AI

May 12, 2025

Published 2 months ago

TL;DR

OpenAI restructures as PBC; Copyright Office rejects fair use for AI training; FDA to fully deploy AI.


Highlights

  • OpenAI and Microsoft are renegotiating their partnership, with Microsoft potentially trading equity for post-2030 tech access; OpenAI plans to restructure its for-profit arm as a public benefit corporation (PBC) to enable new funding and a possible IPO 1.
  • U.S. Copyright Office report rejects fair use for commercial AI training, supporting licensing frameworks and signaling likely regulatory headwinds for AI model developers 2.
  • The FDA will fully integrate AI across all centers by June 30 after a successful AI-assisted scientific review; OpenAI has held meetings with the FDA regarding AI in drug evaluation, and GE HealthCare received FDA clearance for AI-powered imaging tools 3.
  • OpenAI's ChatGPT now reaches 46% usage and powers 70,000+ enterprises; OpenAI appoints Fidji Simo as CEO of Applications to drive expansion into a full AI software suite 6.
  • Microsoft , OpenAI, and AMD urge Congress to boost U.S. AI infrastructure investment to maintain leadership over China, citing compute and energy needs 4.
  • Amazon unveils Vulcan, a tactile AI-powered warehouse robot, and retrains staff as robot technicians, indicating a shift in warehouse automation and labor strategy 8.
  • Meta launches AssetGen 2.0 for generative 3D asset and world creation, rolling out to Horizon developers in 2025 9.
  • Apple to launch AI-powered smart glasses with custom chips in 2027; Meta expands Ray-Ban AI glasses globally and partners with Oakley for next-gen wearables 7.
  • Applied Digital’s 400MW North Dakota AI data center faces customer acquisition challenges amid Nvidia GPU price hikes, Meta’s planned $72B AI infrastructure spend, and surging semiconductor demand 10.
  • Saudi Arabia launches Humain, a multibillion-dollar AI company focused on Arabic LLMs and regional AI ecosystem development 5.
  • Klarna pivots back to human customer service after AI chatbot-led strategy yields lower quality, highlighting limits of automation in customer-facing roles 12.
  • Cybercriminals exploit AI hype to spread malware and deepfakes, with a 220% increase in fraud in Mexico and low user awareness of deepfake risks 15.
  • Carnegie Mellon researchers introduce LegoGPT, an AI model that generates physically stable LEGO structures from text prompts, tested with both robots and humans 13.
  • LTIMindtree secures a $450M, seven-year AI contract with a global agribusiness firm, its largest AI-focused deal to date 14.
  • Elon Musk’s xAI delays Grok 3.5 release by at least a week for further testing 11.

Commentary

OpenAI’s move to restructure its for-profit arm as a PBC, while renegotiating Microsoft ’s long-term stake, signals both a maturing governance model and the need for flexibility to attract future capital 1. The intent to preserve nonprofit board control while enabling new funding and IPO options mirrors Anthropic’s structure and will draw regulatory scrutiny, especially as OpenAI’s enterprise footprint expands rapidly—ChatGPT now powers over 70,000 businesses and processes 100T+ tokens quarterly 6.

Regulatory pressure is mounting: the U.S. Copyright Office’s report explicitly rejects fair use for commercial AI training, backing licensing frameworks and raising compliance stakes for model developers 2. This development, combined with the FDA’s accelerated AI adoption and OpenAI’s involvement in drug evaluation, highlights the increasing intersection of AI with regulated sectors 3. While AI is delivering efficiency gains (e.g., FDA review times reduced from days to minutes), concerns around transparency, bias, and model oversight remain, prompting calls for clear policy guidance 3.

Infrastructure and hardware remain critical battlegrounds. U.S. tech leaders are pushing Congress for greater investment to maintain an edge over China, as compute and energy demands surge 4. Despite bullish forecasts—Meta’s $72B AI capex and McKinsey’s $7.9T data center capex projection—Applied Digital’s difficulties filling a 400MW data center underscore the risks of speculative infrastructure buildouts, especially amid GPU price hikes and supply chain volatility 10.

Product launches and research continue at pace: Amazon ’s Vulcan robot and Meta’s AssetGen 2.0 demonstrate advances in embodied and generative AI 89, while Apple and Meta’s smart glasses signal ongoing convergence of AI and consumer hardware 7. Saudi Arabia’s Humain and LTIMindtree’s $450M AI deal reflect growing global investment and regional ecosystem building 514. Meanwhile, Klarna’s return to human agents after AI setbacks and the spike in AI-enabled cybercrime highlight the operational and security challenges that come with rapid AI deployment 1215.

Subscribe to AI Brief

Get daily ai updates delivered to your inbox