AI Daily — September 9, 2026
2026-09-09
openai.com
OpenAI claims AI-generated solution to Navier–Stokes Millennium Prize Problem
OpenAI has published what it claims is an AI-generated solution to the Navier–Stokes existence and smoothness problem, one of the seven Clay Millennium Prize Problems, including a formal proof verified in Lean. The announcement has been met with significant controversy in the mathematics community, with accusations of procedural irregularities and questions about the validity of the proof. If verified independently, this would represent the most consequential AI scientific achievement to date.
deepmind.google
AlphaGenome Atlas maps molecular effects of 9 billion human DNA variants
Google DeepMind's AlphaGenome Atlas provides predictive models for the molecular consequences of every possible single-nucleotide variant (SNV) across the human genome—roughly 9 billion substitutions. The resource covers regulatory effects, splicing, and gene expression impacts, offering a comprehensive reference for genomics and rare disease research. This extends AlphaGenome's earlier per-query prediction capability into a precomputed genome-wide atlas.
openai.com
GPT-5.6 Sol and Codex used to autonomously run quantum computing experiments
OpenAI showcases an MIT researcher using GPT-5.6 Sol paired with Codex to autonomously execute quantum computing experiments, including qubit calibration and result analysis, with minimal human intervention. The case study demonstrates agentic AI operating in a closed scientific loop—planning experiments, running them on hardware, and iterating based on outcomes. This is a practical demonstration of AI agents accelerating experimental physics workflows.
openai.com
Introducing ChatGPT Images 2.5
OpenAI released ChatGPT Images 2.5, an upgraded image generation system that improves coherence with sketches and reference photos, producing more personalized and stylistically consistent outputs. The update targets creative and professional workflows requiring iterative refinement from existing visual inputs rather than text prompts alone. No technical architecture details were disclosed in the announcement.