Morning Edition · Saturday, June 27, 2026
Tech
OpenAI previews GPT-5.6 Sol, leading with agents, code, biology and cyber
The new top model goes to trusted partners first and is coordinated with the U.S. government, a release pattern as notable as the model itself.
Geopolitics
Anthropic reported near deal to restore models blocked by U.S. export order
The U.S. government suspended access to Fable 5 and Mythos 5 on June 12. The company is now said to be close to lifting those limits through a government agreement.
Tech
DeepSeek open-sources inference optimizations it says speed generation
A new release from the Chinese lab claims generation-speed gains of 60 to 85 percent, the latest in a series of efficiency projects it has published as open code.
Tech
Allen Institute study: hybrid models grasp meaning, transformers copy text
New work from the Allen Institute for AI (Ai2) finds an architectural split, with interleaved recurrent designs better at understanding meaning and pure transformers better at reproducing text word for word.
Tech
EU backs a 400-billion-parameter open model as a sovereignty play
Italy's Domyn, formerly iGenius, will build the model with the EUROPA consortium and European Commission support, with release planned within a year.
Markets
Bank of Korea: AI saves workers 1.5 hours a week, output barely moves
A central-bank survey finds majority adoption and real time savings, but a near-zero link between hours saved and measured productivity.
Tech
Meta ships Segment Anything Model 3 for open-vocabulary perception
The third generation advances text-promptable detection, segmentation and tracking of arbitrary concepts, with early production users already building on it.
Geopolitics
Anthropic proposes rewiring policy for an "AI exponential"
The lab argues that institutions built for slower change cannot match the speed of capability growth, and it offers concrete proposals to narrow the gap.
Tech
Meta details how it builds and tests advanced AI as evals strain to keep up
A new engineering account from Meta's superintelligence effort describes the infrastructure behind scaling, where evaluation is the main obstacle.
Tech
Anthropic launches Claude Tag for team workflows
A new product aims to move Claude from individual assistant to a shared tool embedded in how teams work.
Tech
Google Research presses AI deeper into clinical medicine
The company's Health AI program continues to move models toward real diagnostic and care tasks, a domain where evidence standards are higher than benchmark scores.
Geopolitics
Anthropic's Frontier Red Team and the offensive-cyber question
The lab maintains a dedicated team probing models for dangerous capabilities, as states increasingly consider frontier AI an offensive cyber instrument.