Gemini passes 1 billion users, Nvidia opens financing, Anthropic flags AI spending
Scale, infrastructure, and transparency dominate the day, while new agent models and clinical studies bring practice into focus.
Top stories
Gemini reaches 1 billion users. 63 percent use voice. Google reports 150 million images generated per day. ChatGPT is also above 1 billion users.
Nvidia wants to secure the value of its own chips. Financial partners are to mobilize up to 500 billion dollars for AI infrastructure. That lowers capital costs for data centers.
Anthropic rents data center capacity worth 9.1 billion dollars from Riot Platforms. That secures additional GPU resources. The focus is on scaling Claude services.
Anthropic introduces invisible watermarks for Claude. Watermarks—machine-readable markers—are to be embedded in texts and images. Supported files carry signed provenance data and target EU transparency.
Researchers reveal an API vulnerability at several AI providers. Hidden chain-of-thought and sensitive strings like passwords could be extracted. The case highlights risks in prompt and output channels.
Tools and releases
- NVIDIA Nemotron 3.5 Lightning, an open Mixture of Experts model—a model type with specialized subnetworks—is optimized for always-on agents and is available in Ollama.
- OpenAI introduces 125-dollar Premium Seats for ChatGPT Business to handle agentic workloads with high token demand.
- OpenAI tests ads in ChatGPT with clear labeling and user controls.
- Daybreak security models from OpenAI are available via AWS Bedrock and target enterprise workflows.
- Mistral launches in-region inference and expands European infrastructure for sovereign AI.
Research
- AMIE shows real-time video consultations in an initial study and demonstrates clinical suitability in doctor–patient conversations.
- CARE X advances VLMs—vision-language models—with auxiliary supervision, reward-based alignment, and measurement tools for improved chest X-ray interpretation.