GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
THE SO WHAT
If GPT‑5.6 Sol Ultra can generate a credible proof of the Cycle Double Cover Conjecture, frontier models are crossing from pattern-matching into territory that challenges how we do formal reasoning and verification. For technical orgs, the bottleneck shifts to building tooling and processes that can audit machine-generated theory at scale before it flows into systems and standards.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIMeta’s stock roars back to life as it heads for its best week in years
Public markets are rewarding Meta’s low-cost AI infra and pricing story — investors are buying the idea that scale plus cheaper inference can defend margins. If you’re competing on AI infra or platform economics, assume customer expectations on price/performance just ratcheted down again.
Applied AIThe Department of Commerce loosens export controls to the UAE, letting G42 and US companies like Apple, Meta, and xAI export AI chips to UAE without a license
Looser AI chip export controls to the UAE open another high-capacity region for training and inference — and formalize G42 as a central node in that ecosystem. If you’re planning global AI infra, the Gulf just became a more straightforward region for serious deployments, but you still need a clear view on data residency and partnership risk.
Applied AIWall Street is debating the AI buildout. Enterprises just answered: 86% say their GPUs run at half capacity or less
GPU underutilization at 50% or worse means the constraint isn’t hardware anymore — it’s orchestration, workload design, and governance. If your board is asking for more GPUs, first instrument current clusters and kill stranded capacity before you add another dollar of capex or long-term commitments.
Applied AIAI Labels Could Be Coming to Music Streaming Platforms Soon
If streaming platforms add AI labels, they’re turning provenance into a UX and policy surface — not just a backend metadata field. Any brand or creator using generative tools should assume disclosure becomes default and design content strategies that work in a labeled world, not hope to blend in.