Morning Edition · Sunday, June 28, 2026
Tech
AI's First Revenue Census Puts Generative-AI Sales at $110 Billion, Above Infrastructure Depreciation
Exponential View's first report measures real end-customer spending, and it finds that quarterly sales outside China now exceed the quarterly cost of data centers and chips wearing out, the first time that has happened.
Geopolitics
Washington Orders Anthropic to Cut Foreign Access to Two Top Models, Citing a Narrow Jailbreak
An export-control directive required Anthropic to disable its Fable 5 and Mythos 5 models worldwide for all foreign nationals, extending government control from computer chips to access to the models themselves.
Tech
DeepSeek Open-Sources DSpark, Claiming Up to 4x Throughput on V4 via Speculative Decoding
The semi-parallel decoding method ships as an add-on module to the existing V4 Flash and Pro checkpoints, and DeepSeek also released code for training draft models on other architectures.
Tech
MirrorCode Benchmark Finds Frontier Models Can Rebuild Real Software From Scratch, Up to a Point
Epoch AI and METR ask models to rebuild working programs without any source code, and the best result is a 56 percent solve rate on tasks estimated to take human engineers weeks.
Tech
EU Picks an Italian Startup to Build a 400-Billion-Parameter Open Model in 24 Languages
The European Commission named the Domyn-led EUROPA consortium as the sole winner of its Frontier AI Grand Challenge, granting it up to 2.5 percent of the European Union (EU) supercomputing network EuroHPC for a year.
Tech
Open-Vocabulary Segmentation Reaches Production as Apps Build on Meta's SAM 3
Meta's promptable concept-segmentation model, which finds and tracks any specified object from a text prompt, is moving from benchmark tests into shipping consumer products.
Tech
Anthropic Asks Washington for Power to Block Dangerous Models, Plus $350 Million for Labor Fallout
The lab's "Policy on the AI Exponential" proposes oversight modeled on the Federal Aviation Administration (FAA), with mandatory third-party testing above a compute threshold, days before the government pulled two of its own models under an export order.
Geopolitics
Alibaba Sues the Pentagon to Get Off Its Chinese-Military-Company List
The challenge to the Section 1260H designation adds a legal dimension to the contest between the United States and China over who controls artificial intelligence and cloud computing.
Tech
Anthropic Puts Claude Inside Slack as a Taggable Teammate With Shared Memory
Claude Tag gives a channel one shared agent that anyone can hand work to. It builds context over time and can act on its own, a clear move to put agents into everyday workflows.
Markets
A Venture Investor Accuses AI Leaders of Using Doom Warnings as a Fundraising Tool
Chamath Palihapitiya's claim that warnings about existential risk lined up with funding rounds is unproven, but it reflects a genuine shift in how the market interprets safety rhetoric.
Tech
Brain-to-Image Decoding Advances, and DeepMind's Hassabis Frames It as AI for Science
Researchers are using diffusion models to reconstruct images people have seen from functional magnetic resonance imaging (fMRI) brain scans, and newer work is attempting to influence mental imagery and dreams through language.
Tech
Meta's Muse Spark Shows the Efficiency Turn in Frontier Models, and a Break From Open Weights
Meta Superintelligence Labs says its first model matches the capability of Llama 4 Maverick at over an order of magnitude less compute, and unlike Llama it is closed.