Secretary of War Tweets That Anthropic is Now a Supply Chain Risk
Don't Worry About the Vase Podcast
Practicality of Technical Safeguards Against Misuse
Zvi questions whether technical measures can reliably stop lawful-but-objectionable surveillance or jailbreaks.
The Don’t Worry About the Vase Podcast is listener-supported. To receive new posts and support ongoing production, consider becoming a free or paid subscriber.
This is, as far as I can recall, the longest converted Don’t Worry About the Vase post, clocking in at three hours and seven minutes. If you found it valuable, paid subscriptions really do help cover the premium ElevenLabs voices used for maximum reliability, variety of speakers and audio quality.
* 00:00:00 - Introduction
* 00:00:44 - What Happened
* 00:10:16 - The Timeline Of Events
* 00:22:48 - I Did Not Have Time To Write You A Short One
* 00:23:47 - The Unhinged Declaration of the Secretary of War
* 00:26:15 - Altman Has Been Excellent On The Question of Supply Chain Risk, But May Need To Do More
* 00:28:55 - Arrogance Here Means Insisting On Meaningful Red Lines On Mass Domestic Surveillance and Lethal Autonomous Weapons
* 00:31:03 - Not Doing Business Is Totally Fine
* 00:31:46 - The Demand For Unrestricted Access Is New And Is Selective And Fake
* 00:34:15 - Claims Of Strongarming Are Ad Hominem Bad Faith Obvious Nonsense
* 00:36:26 - Hegseth Equates Not Being a Dictator With Companies Having Veto Power Over Operational Military Decisions
* 00:42:17 - The Part That If Enacted Would Be A Historically Epic Clusterfuck
* 00:50:44 - The Other Part Of The Clusterfuck
* 00:54:47 - The Department of War Had Many Excellent Options
* 00:57:54 - And Then There’s Emil Michael
* 01:04:52 - Anthropic Will Probably Survive
* 01:07:29 - The Goal of DoW Was Largely Mass Domestic Surveillance
* 01:18:16 - What Are The Key Differences Between The Two Contracts?
* 01:26:16 - OpenAI’s Contract Terms
* 01:30:01 - What OpenAI’s Contract Terms Actually Do
* 01:32:23 - OpenAI Is Trusting DoW And Sam Altman Misrepresented This
* 01:36:29 - OpenAI Accepted Terms Anthropic Explicitly Declined And That Would Not Have Protected Anthropic’s Red Lines
* 01:38:47 - How Altman Initially Described His Deal
* 01:45:45 - OpenAI Allowed All Lawful Use And Trusts DoW On This
* 01:51:05 - The DoW Could Alter This Deal
* 01:53:18 - Why OpenAI’s Shared Legal Language Offers Almost No Protections
* 02:03:48 - So How Does OpenAI Hope For This To Work Out?
* 02:06:44 - This Was Never About Money
* 02:10:06 - OpenAI Tells Us How They Really Feel
* 02:11:52 - First The Good News
* 02:16:53 - The OpenAI Redlines Only Forbid Currently Illegal Activity
* 02:22:01 - Altman Does Not Present As Understanding The Difference In Redlines
* 02:24:25 - Meeting Of The Minds
* 02:27:11 - Anthropic’s Position Was The Opposite Of How This Is Portrayed
* 02:27:45 - The Room Where It Happened
* 02:31:36 - You Don’t Have The Right
* 02:38:12 - I Ask Questions And Get Answers
* 02:43:05 - Does This Contract Apply To NSA?
* 02:44:10 - Can OpenAI Models Be Used To Analyze Commercially Available Data At Scale?
* 03:02:22 - Employee Activism
Get full access to DWAtV Podcast at dwatvpodcast.substack.com/subscribe
The AI-powered Podcast Player
