Morning Edition · Sunday, June 21, 2026
Geopolitics
Washington Turns Export Controls on a Model Itself, Forcing Anthropic to Disable Fable 5 and Mythos 5
A national-security directive bars foreign access to two frontier models, a shift in which the United States treats model access, not chips or weights, as the controlled item.
Geopolitics
The NSA, Mythos, and the Quiet Arrival of an AI Cyber Doctrine
Reporting that an Anthropic model breached classified networks within hours has become a reference point for how fast offensive AI capability is advancing beyond oversight.
Tech
Amodei Asks Government for the Power to Block Unsafe Models
Anthropic's "Policy on the AI Exponential" proposes binding rules tied to a compute threshold, a notable step from a frontier lab that is still releasing frontier models.
Markets
Nvidia Raises 25 Billion Dollars in Bonds to Fund the AI Buildout
The chipmaker's first debt sale since 2021 drew 85 billion dollars in orders, as Jensen Huang describes AI as a structure that begins with energy and a competition without end.
Tech
Meta's SAM 3 Pushes Open-Vocabulary Segmentation Toward Production
The model can segment every matching instance from a text prompt, and a SAM 3.1 update aims at faster real-time video tracking. Together they move promptable perception past fixed label sets.
Tech
Genesis AI Unveils Eno, a Wheeled Robot Built on a Robotics Foundation Model
The startup avoids humanoid imitation in favor of a mobile manipulator driven by its GENE foundation model, with industrial deployments targeted by the end of 2026.
Tech
Meta's Muse Spark Marks a Break From Llama, and From Open Weights
Superintelligence Labs' first model is natively multimodal with a parallel sub-agent reasoning mode, and it is closed source.
Tech
OpenAI Hires Trump AI Plan's Architect to Run a New Strategic Futures Team
Dean Ball, primary author of the administration's AI Action Plan, joins OpenAI on July 6 to build frontier policy and engage Congress.
Tech
OpenAI Adds Record-and-Replay Automation to Codex on macOS
A desktop feature records repetitive user actions and converts them into structured, replayable scripts, moving agents further into routine workflows.
Tech
Google's Earth and Health AI Show the Slow March Into Real-World Science
Domain foundation models for planetary sensing and clinical tasks advance the case for AI in scientific discovery, with the caveat that peer-reviewed evidence lags behind the demonstrations.
World
Anthropic Launches Claude Corps as Labs Court Social License
A national fellowship to extend AI's benefits arrives as public resistance to AI broadens beyond disputes over data centers.
Tech
Meta Details How It Builds and Tests Advanced AI as Evaluation Becomes the Bottleneck
A look inside the lab's evaluation and infrastructure practices comes as the field acknowledges that testing, not training, increasingly determines what can be deployed.