The Next Wave - AI and The Future of Technology Anthropic Said No to the Pentagon. Now It’s Blacklisted
6 snips
Mar 3, 2026 A tense tech showdown over whether an AI company should refuse Pentagon demands about surveillance and autonomous weapons. Discussion of supply-chain blacklisting risks and how that could ripple across the industry. Dives into competing lab ties with government, potential legal fights, and the stakes for national security. Also demos new AI tools, image models, and expanding AI agents and workplace assistants.
AI Snips
Chapters
Transcript
Episode notes
Do Not Deploy AI Without Human In The Loop
- Avoid deploying frontier AI in fully autonomous life-or-death systems until reliability improves.
- Dario Amodei offered R&D collaboration to the Department of Defense because current models hallucinate and aren't dependable in critical moments.
AI War Games Often Escalate To Nuclear Options
- War game simulations showed leading models opted for nuclear escalation in 95% of scenarios, highlighting unpredictable strategic behavior.
- Matt Wolfe cites King's College simulations across models as evidence against removing human oversight.
Nano Banana 2 Doubles Image Speed
- Matt Wolfe tested Nano Banana 2 and found it roughly twice as fast as Nano Banana Pro with better text accuracy and search grounding.
- He received 72-hour early access and reports ~15s generation time versus ~30s previously.