Morning Edition · Thursday, October 8, 2026
Tech
OpenAI's 722 machine-written math manuscripts land on mathematicians to verify as Terence Tao warns of a rush to "flatten" open problems
According to the repository's own catalogue, Lean formalizations cover the main result of only 162 of the 722 papers, and the Clay Mathematics Institute still lists Navier-Stokes as an open problem.
Tech
Anthropic prices Claude Haiku 5.5 at $0.10 per million input tokens, exactly matching OpenAI's GPT-6 Luna
With Sonnet 5.5 and GPT-6 Sol both listed at $2 and $10 per million tokens, Anthropic and OpenAI now charge identical prices at two tiers, and both small models list below DeepSeek V4.1 Flash's off-peak rate.
Tech
GPT-6 reaches all ChatGPT users with interactive "Intelligent UI" as OpenAI adds 512 MB audio uploads for paid plans
Free and Go users receive GPT-6 Luna starting today, and OpenAI says GPT-6 Instant starts answering web-search queries 44% sooner than GPT-5.6 Instant, a figure that measures time to first token and not total answer time.
Tech
Nvidia and Microsoft open pre-orders for Arm-based RTX Spark Windows laptops with up to 128 GB of unified memory
The Surface Laptop Ultra starts at $2,599 with 24 GB of memory, ships October 16, and costs $5,899.99 in the top 128 GB configuration, which is large enough to hold mid-sized open-weight models locally.
Tech
White House gathers more than $1 billion in private pledges from AMD, OpenAI and Anthropic for its Genesis Mission science program
The administration says total commitments across the mission exceed $5 billion as it relies on industry compute to replace federal science funding it has cut.
Tech
Google and Cambridge's VeriHarness uses the generating model to audit its own agent rollouts and adds about 6 points on long-horizon tasks
The method treats agreement between rollouts as suspect and disagreement as a clue, and the authors have released the code with about 26,000 agent traces.
Tech
Google releases EmbeddingGemma 2, a 740-million-parameter open embedder for text, code, images and audio under Apache 2.0
Google reports that the text-only configuration runs in about 191 MB of active memory on a quantized Pixel 11 Pro, and that its score on the Massive Text Embedding Benchmark's code track rose nearly 10 points over the first version.
Tech
MIT's open-weight FleXray segments 60 anatomical structures in X-rays after training only on synthetic images rendered from CT scans
The team skipped manual X-ray annotation and used a physics-based engine to generate fully labeled 2D X-rays from existing 3D whole-body segmentations.
Macro
Anthropic's economics team models US GDP in 2030 at 1.6% to 32.4% above a no-AI baseline, with labor's share of income falling to as low as 45%
In the most extreme of its three scenarios unemployment approaches 12%, and wages for physical occupations rise in every scenario while screen-based knowledge work takes the losses.
Tech
Two new papers show standard metrics hiding real costs, from tokenizer penalties on non-Latin scripts to a diarizer that worsens while appearing to improve
Tokka-Bench compares seven tokenizers across 100 natural and 20 programming languages and finds that how a vocabulary is allocated matters more than its size.
Tech
New security papers show smart bulbs leaking data through light and reinforcement-learning attackers learning to evade edge intrusion detection
A companion study gives a quantitative framework for deciding where machine-learning intrusion detection can run in Internet of Things deployments.
Tech
"Building Rome from a Single Image" generates complete 3D scene meshes, including unseen surfaces, from one photograph
The method uses distance-adaptive chunks and explicit labeling of observed versus unobserved regions, and it adds about 4,000 synthetic outdoor scenes to fix an indoor-heavy training data supply.