← All episodesEP 28

Is Claude Becoming the Government's Voice?!

October 20, 2026 · 90 min

EP 28October 20, 2026 · 90 minShip It or Skip It

Claude may soon answer U.S. voters' questions with state-specific information from election sources. Anthropic says the direct-answer feature is not live yet. Tesla's next-generation Roadster briefly lifted and rotated in a demonstration, but that does not make it a practical flying car.

In Episode 28 of Human in the Loop, Oscar Gallo and Matt Wozniak ask what a product announcement or demo actually proves.

Watch on YouTubeListen on SpotifyListen on Apple
Signal or Noise

The week’s AI headlines, filtered.

  1. SIGNALClaude Voting Info

    Anthropic's planned voter-information answers and the need to check source, location, and freshness.

  2. SIGNALTesla Roadster Jump & Rotate

    Tesla's Roadster demonstration. A striking prototype is not the same as a delivered, road-approved passenger vehicle.

  3. SIGNALAnthropic Warns of AI Self-Replication

    Anthropic's experiments on model self-replication. The report raises a security question, not evidence of an AI worm currently spreading online.

  4. SIGNALBank of America AI Vulnerability Remediation

    Bank of America's reported use of about 1,000 contractors for AI-linked vulnerability remediation, alongside roughly 3,000 existing personnel. The figures do not prove AI discovery outpaced remediation or that all contractors worked only on Mythos findings; reported remediation costs also included separate compliance work.

  5. SIGNALAnthropic Reports Decline in Junior Hires

    Anthropic's reported decline in the share of junior technical hires. Company-specific figures, not an industry-wide forecast.

Ship It or Skip It

Two checks on AI-generated code.

Mutation Testing

Mutation testing as a tougher check for tests around AI-generated code.

AI Code Review

AI code review as a second pass on generated changes, with a human responsible for the merge decision. No vendor or performance result is claimed.

Unpopular opinions

Two opinions, no disclaimers.

Oscar

“I don't want opinions from my AI. It shouldn't have opinions on politics, movements, nothing. A model should be a logic machine that gives you the information you need and offers no personal opinions, because it shouldn't have any.”

Matt

“An AI assistant should never speak in the government's voice. Point, don't pronounce: give me the election office, the link, and the date it was last updated, and let the human be the human in the loop.”

Sources referenced