Above the Law - Thinking Like a Lawyer At Least The Robots In The Coming War Against Humans Will Understand War Crimes
9 snips
Dec 17, 2025 The Pentagon's new AI bot shockingly flags potential war crimes, showcasing its capabilities over human lawyers. The hosts debate the complexities of regulating large language models and the risks of bias in militarized AI. They discuss an Executive Order that challenges state AI regulations and ponder on liability for AI’s discriminatory outputs. In the legal arena, Justice Sotomayor's strategic questioning at the Supreme Court raises eyebrows amid concerns over the influence of liberal justices. Finally, the episode hints at the changing dynamics of court politics.
AI Snips
Chapters
Transcript
Episode notes
Grinch At A Firehouse Party
- Chris Williams recounted his cousin's firehouse Christmas party with Santa and the Grinch.
- He described kids chasing the Grinch and a cute apology moment that delighted his daughter.
Military AI Outpaces Human Commanders
- The Pentagon's new internal AI flagged an example order as a war crime faster than human commanders did.
- Joe Patrice argues bots sometimes outperform senior military judgment on legal norms.
Scaling Makes Guardrails Slippery
- Large language models resist fixed guardrails as they scale because they optimize next-word prediction.
- Joe Patrice warns that bigger models can circumvent intended safety controls.