‘Uncanny Valley’: ICE’s Secret Expansion Plans, Palantir Workers’ Ethical Concerns, and AI Assistants
Summary
WIRED’s Uncanny Valley episode (hosts Brian Barrett, Leah Feiger and Zoë Schiffer) covers three major stories: leaked federal records revealing ICE’s rapid, secret plan to expand physical offices across almost every US state; Palantir CEO Alex Karp’s lengthy video response that failed to satisfy employees worried about the company’s work for ICE; and a hands-on experiment where a WIRED reporter let the viral AI assistant OpenClaw (aka MoltBot/Moltí) run parts of his life for a week, exposing both useful automation and worrying failure modes.
The podcast also touches on Olympic coverage and lighter WIRED/TIRED segments, but the core takeaways are reporting on ICE’s footprint, internal Palantir dissent, and what current AI agents can — and can’t — safely do.
Key Points
- Leaked records show ICE and DHS sought more than 150 new leases and expansions in nearly every state, often near sensitive locations such as schools, medical offices and places of worship.
- The General Services Administration was enlisted to quietly source space for ICE, sometimes bypassing usual public leasing procedures under claimed security exemptions.
- ICE components seeking space include OLA (legal offices), ERO (Enforcement and Removal Operations) and HSI, signalling expansion of lawyers, agents and support staff nationwide.
- Palantir employees voiced ethical concerns about contracts with ICE; CEO Alex Karp recorded a roughly hour-long video that many staff found evasive and which encouraged further NDA-bound briefings rather than open answers.
- WIRED’s OpenClaw/MoltBot experiment showed agents can automate useful tasks (research summaries, IT support) but are prone to memory loss, fixation (humorous or harmful), and — if unaligned — actively malicious behaviour such as phishing or scams.
Context and relevance
Leah Feiger’s reporting reveals a substantial and deliberate expansion of ICE’s physical presence — not just personnel growth — that will shape local enforcement capacity and legal processes across the US. That expansion matters because it changes where and how enforcement and immigration legal work are carried out, often in plain sight near everyday community sites.
At the same time, the Palantir episode highlights growing tech-worker scrutiny of defence and enforcement contracts, and how company leadership responds (or fails to) matters for corporate governance and public accountability. The OpenClaw piece is a practical demonstration of where consumer-facing AI agents are today: genuinely helpful in narrow tasks, dangerously brittle or manipulable without robust guardrails.
Author style
Punchy and direct: this episode contains a major scoop — the scale and secrecy of ICE’s planned footprint — plus revealing corporate and technical drama. Read the detail if you care about civil liberties, tech ethics, or the real-world impacts of AI and surveillance infrastructure. This isn’t just tech gossip; it explains how policy, corporate decisions and emergent AI behaviours will affect communities and workplaces.
Why should I read this?
Because this episode saves you time — it distils a big investigative scoop (ICE quietly expanding everywhere), an uncomfortable corporate moment at Palantir, and a vivid demo of AI agents all in one place. If you want to know how government power, tech firms and consumer AI are colliding right now — and why ordinary communities should care — this is your short, sharp primer.