I Let a Local AI Read My Ballot—Here's What Happened
Tim and Paul break down this week's chaotic AI news, starting with Tim's experiment using a locally hosted model to research his ballot—steelmanning both sides and checking every claim against source documents. They also cover the quarter-billion-dollar AI and crypto PAC losing streak, frontier models giving opposing answers across nearly 500 midterm races, Postman quietly installing itself into coding tools, Google training on paid users' chats, a $9,400 consumer GPU, Sam Altman's controversial take on your RAM, and Apple's foldable iPhone Duo cracking within hours. It's a practical look at who shapes the information you vote with and the tools you use.
28 Oct 2026
AI Is Outrunning the Systems That Check It
Tim and Paul dig into Anthropic's IPO filing and the uncomfortable question it raises: should a civilization-scale AI lab even be a corporation when its own risk factors include nationalization and mandatory shutdown? They trace the Manhattan Project comparison and then zoom out to a bigger industry bottleneck—AI now builds faster than we can verify, from AWS's race-condition outage to an agent that deleted a production environment—and Tim shares two practical experiments for reclaiming control: a software factory and spec-driven development. This week's news roundup covers Gemini 4 Argon, TypeSafe's image-reading decision model, and Anthropic's multi-agent code review that puts four agents on every pull request.
23 Oct 2026
Who's Responsible When AI Moves Faster?
In this episode, Tim and Paul break down a week where AI sprinted past its safeguards: Google's Gemini 3.8 Flash ships to serious skepticism, two critical PHP engine vulnerabilities found by AI send patching teams into overtime, and major banks quietly detect unauthorized AI agents acting inside customer accounts while executives warn of 'liability transfer.' They also unpack Equifax's eyebrow-raising claim that your credit file is an 'AI data moat,' and end with Flint, a new open-weight coding model that runs on a laptop. It's a candid look at who's really accountable when AI moves faster than the systems we built to watch it.
Host of Rubber Duck Radio?
Claim your podcast to manage its listing and keep your show details accurate.
Pod Engine is an independent podcast discovery and analytics service and is not affiliated with or endorsed by this podcast. Artwork and show content belong to their owners. Full legal notice.
