Hard Fork

Tech Grapples With ICE + Casey Tries Clawdbot, a Risky New A.I. Assistant + HatGPT

with Kevin Roose and Casey Newton
30 Jan 2026 8 min read 45m
TL;DR

Tech companies face mounting pressure over immigration enforcement partnerships, with major platforms reconsidering their relationships with government agencies. The episode explores how AI assistants are becoming riskier tools with real-world consequences, particularly as companies race to deploy new features without fully understanding the implications.

Key Moments
[No transcript — approximate]
“[No transcript — approximate] Minneapolis has become a battleground where the choice of phone technology determines how people interact with immigration enforcement”
Introduction to the episode's opening story about ICE technology impact
[No transcript — approximate]
“[No transcript — approximate] Tech companies are realizing their infrastructure is being used in ways they didn't anticipate or approve of”
Discussion of unintended consequences of tech platform partnerships
[No transcript — approximate]
“[No transcript — approximate] Clawdbot represents the kind of experimental AI that moves fast without enough safety testing”
Casey's experience testing the new AI assistant tool
[No transcript — approximate]
“[No transcript — approximate] The real risk isn't the AI itself—it's that people trust it too much before we understand what it can do”
Analysis of why new AI assistants pose unique dangers
[No transcript — approximate]
“[No transcript — approximate] HatGPT shows how quickly AI gets weaponized for niche but problematic use cases”
Broader pattern of specialized AI tools with harmful applications
About the show
›
Hard Fork is the New York Times podcast where Kevin Roose and Casey Newton dig into the week's biggest tech stories. The show examines how technology is reshaping society, from AI developments to platform politics to the human impact of innovation. With sharp analysis and accessible explanations, Hard Fork helps listeners understand what tech companies are really doing and why it matters.
Takeaways
1
Tech's ICE problem isn't going away As immigration enforcement becomes more tech-dependent, platforms face pressure to audit and potentially sever partnerships with government agencies. The Minneapolis case shows how infrastructure choices—from phone networks to data access—have direct human consequences, pushing companies to take public stances on controversial partnerships.
2
Move-fast culture creates AI trust gaps New AI assistants like Clawdbot are being released before their failure modes are fully understood, and users don't know when to trust them. The gap between capability and safety understanding creates real risks when people rely on half-tested tools for important decisions.
3
Specialized AI tools enable targeted harm Projects like HatGPT demonstrate that AI capabilities get rapidly applied to niche, harmful purposes once the underlying models are released. The problem isn't just general-purpose AI—it's how easily specialized versions get built and deployed without clear governance frameworks.