Taylor Lorenzâs Power User Inside The AI Scandal Rocking Silicon Valley: How The Govt Wants AI To Kill Without Humans
34 snips
Mar 6, 2026 Ross Anderson, Staff writer at The Atlantic who investigates AI, national security, and surveillance. He unpacks the Pentagonâs showdown with AI firms over autonomous weapons and mass surveillance. Short, sharp stories about a $200M contract fight, Anthropic pushing back on lethal uses, OpenAI stepping in, and why commercial data access sparked a national controversy.
AI Snips
Chapters
Transcript
Episode notes
Why The Pentagon Picked Claude First
- The Pentagon awarded Anthropic a $200M contract because Claude was seen as the safest and most reliable model for classified systems.
- Anthropic obtained early government clearance and earned commander-level praise, which made its model integral to classified workflows within months.
How A Classified Raid Sparked The Blowup
- The rupture began after Anthropic learned its tech had been used in an operation tied to Venezuela, prompting executives to confront the Pentagon.
- Pete Hegseth's aggressive response labeled Anthropic a supply chain risk, escalating the dispute into a political showdown.
Cloud Only Is Not A Real Barrier To Killer Drones
- The core technical dispute was whether 'cloud-only' usage prevents autonomous weapons deployment.
- Anthropic argued cloud vs edge is a false distinction because mesh networks and edge computing push cloud control into the field.