Morning Edition · Monday, June 29, 2026
Markets
Oracle Puts AI-Driven Job Cuts in a Securities Filing, a First for a Company Its Size
The database and cloud company tied part of a roughly 21,000-person reduction directly to AI adoption, even as its data-center capital spending reached $55.7 billion.
Geopolitics
Washington Reopens Anthropic's Mythos Model to More Than 100 Trusted United States Firms
Two weeks after ordering Anthropic to disable its most capable model for foreign nationals, the Commerce Department cleared access for vetted American institutions.
Tech
A Chinese Open-Weight Model Is Reported to Match Claude on Cybersecurity Tests
Independent benchmarking found Zhipu's GLM 5.2 outscoring Claude Code on one vulnerability-detection task at much lower cost, days after Washington restricted access to comparable United States models.
Tech
xAI Says Grok 4.5 Runs on a 1.5-Trillion-Parameter Foundation, but Only in Private Beta
Elon Musk announced the model on his social platform with internal evaluations and no public access, citing training data drawn in part from Cursor.
Markets
United States Data-Center Construction Spending Passes Transportation Infrastructure
Outlays exceeded $50 billion and, by one measure, surpassed combined public spending on airports, ports and mass transit.
Tech
A New Paper Argues Prompt Injection Cannot Be Fully Defended in Today's Architectures
Researchers offer a proof that models which share one embedding space for instructions and data lack the separation any robust defense would require.
Tech
Researchers Propose Axioms to Test What a Model's Latent Reasoning Actually Represents
An axiomatic evaluation framework aims to surface representational failures that benchmark accuracy can hide.
Tech
A Training Paradigm Tries to Make LLM Agents Plan Ahead Instead of React
The proposed method internalizes a world model so agents can evaluate hypothetical futures during long-horizon tasks.
Tech
Google Makes Its AI Coding Team Permanent and Pushes Into Midtraining
DeepMind is broadening a months-old coding effort to work between pretraining and post-training as it tries to match Anthropic in coding.
Tech
Meta's Segment Anything 3 Pushes Open-Vocabulary Perception Toward Production
The latest version of the segmentation model is being used in shipping consumer products, a sign promptable vision is moving past research demos.
World
OpenAI Maps Which European Jobs AI Will Change, Grow or Automate
A new report from the lab estimates how AI could reshape occupations across the European Union, a deliberate move into labor-market analysis.
Geopolitics
Anthropic Presses Lawmakers to Rebuild Institutions for an AI 'Exponential'
The lab is circulating policy proposals arguing that the pace of AI progress has moved faster than a policymaking process built for a slower era.