Marketplace Tech Bytes: Week in Review — Anthropic's new AI model, a referendum on data centers, and NASA livestreams journey to space
7 snips
Apr 10, 2026 Joanna Stern, founder of New Things and tech columnist, breaks down big tech and space stories. She unpacks Anthropic’s Claude Mythos Preview and its limited Project Glasswing rollout. She discusses a Wisconsin referendum slowing data center growth and the environmental debates around them. She also covers Artemis II’s viral NASA livestreams and how astronauts use social media to reconnect the public with space.
AI Snips
Chapters
Transcript
Episode notes
Powerful AI Models Need Controlled Release
- Anthropic's Claude Mythos Preview excels at finding long-standing security exploits but is considered too dangerous for public release.
- Anthropic restricts access to 40+ companies via Project Glasswing to let trusted organizations patch vulnerabilities first.
AI Companies Preemptively Limit Model Access
- AI firms are increasingly self-editing, withholding models to prevent misuse like mass surveillance or weapons creation.
- Joanna Stern notes firms will route models first to security researchers and partner companies before public release.
Cybersecurity Roles Will Become AI Management
- AI will reshape cybersecurity roles into managers of AI agents rather than pure manual vulnerability hunters.
- Joanna Stern compares this shift to how coding and customer service now involve directing AI tools instead of doing all tasks by hand.