The Powers That Be: Daily Trump’s A.I. Army
5 snips
Mar 6, 2026 Ian Kreitzberg, a journalist and author covering AI policy and industry disputes. He unpacks the Pentagon’s shifting use of AI in combat, intelligence, and surveillance. He traces the Anthropic conflict over guardrails, reports about model use in Iran, and how OpenAI moved in amid the fallout. Short, sharp takes on contracts, public messaging, and political tensions around AI.
AI Snips
Chapters
Transcript
Episode notes
AI CEOs Acknowledge Technical Limits
- Anthropic publicly set red lines against autonomous weapons and domestic surveillance, framing safety as both ethical and technical concerns.
- Dario Amodei argued models aren't capable yet for autonomous warfare, and Sam Altman later echoed capability limits and need for human safety engineers.
Public Fallout After Trump Ordered Agencies To Drop Anthropic
- The dispute exploded after Trump told federal agencies to stop using Anthropic and Pete Hegseth labeled it a supply chain risk.
- Reporting then revealed Claude was used via Palantir for real-time mapping in Iran, intensifying the controversy.
Require Human Safety Stacks Before Military Use
- Treat claims about AI readiness skeptically and demand safety stacks with human oversight before military deployment.
- Experts caution LLMs are black boxes with bias, cybersecurity risks, and unreliable outputs unsuitable for targeting.