Tech Brew Ride Home – January 26, 2026

Episode Title: Now We Have The Clawdbot
Host: Brian McCullough
Duration: ~15 min

Episode Overview

Brian McCullough recaps the top tech stories of the day, focusing on evolving privacy policies, hardware launches, regulatory scrutiny of AI, new risks associated with cloud-based encryption key storage, and a deep-dive into the open-source "Claudebot" project—a local AI assistant pushing the envelope for personal productivity and digital autonomy.

Key Discussion Points & Insights

1. TikTok’s New US Privacy Policy and Ownership (00:35–04:25)

• Theme: TikTok, transitioning ownership to a US-majority joint venture as required by US law, launches an updated privacy policy for American users.
• Key Changes:
  • Precise Location Tracking: Now collects GPS-level location data with user permission, bringing it in line with Instagram and X.
  • AI Interactions Logging: Explicitly collects and stores prompts and outputs from all AI tool interactions, including attached metadata.
  • Expanded Ads Network: Broader sharing of user data externally, meaning ad targeting may follow users beyond the TikTok ecosystem.
• Memorable Quote:

  “It’s easy to tap agree and keep on scrolling through videos on TikTok, so users might not fully understand the extent of changes they are agreeing to with this popup.” (Brian McCullough, 01:26)

2. Apple’s Second-Generation AirTag (04:25–06:00)

• Highlights:
  • Upgraded with Apple’s second-gen ultra-wideband chip (as seen in iPhone 17, new Apple Watch models).
  • Improved Features:
    • 50% greater precision and range for locating items.
    • Speaker 50% louder—better for finding lost objects.
    • Privacy/security upgrades described as “industry first protections against unwanted tracking,” with details kept vague intentionally for security.
  • Integrated with Apple Watch; users can now use their watch alone for “precision finding.”
• Quote:

  “The new Airtag is 50% louder than the previous generation, enabling users to hear their AirTag from up to 2x farther away than before.” (From 9to5Mac/Apple newsroom, read by Brian, 05:32)

3. EU Investigates XAI Over ‘Grok’ Deepfake Controversy (06:01–07:08)

• Story:
  The European Union opens a formal Digital Services Act (DSA) investigation into XAI after Grok LLM generated sexualized deepfakes, including of women and children.
• Risks:
  • Potential for fines up to 6% of global revenue.
  • Key investigation point: Whether XAI took sufficient preventative measures to stop abuse.
• Standout Quote:

  “Non-consensual sexual deepfakes of women and children are a violent, unacceptable form of degradation.” (EU tech chief Hanna Verkunen, 06:24)

4. ChatGPT’s GPT-5.2 Citing ‘Grokipedia’: Disinformation Risks (07:09–09:34)

• Story:
  Tests showed GPT-5.2, when fielded via ChatGPT, sometimes cites Grokipedia—a questionable, possibly malign source—in responses to varied queries.
• Background:
  • Disinformation experts raise alarms about the risk of LLM “grooming,” where inaccurate data is seeded into AI datasets deliberately by propaganda actors.
  • Problem: If LLMs cite unreliable sources, those sources seem legitimate to end users.
• Notable Perspective:

  “Grokipedia entries … are relying on sources that are untrustworthy at best, poorly sourced and deliberate disinformation at worst.”
  —Nina Jankowicz, disinformation researcher (08:46)

  • Removing entrenched falsehoods from AI responses is extremely difficult, even after corrections are published.
• Warning:

  “Most people won’t do the work necessary to figure out where the truth actually lies.”
  (Nina Jankowicz, 09:23)

5. Microsoft BitLocker Key Disclosures (11:50–13:35)

• News:
  Microsoft confirmed it provides BitLocker recovery keys to law enforcement with a valid legal order—but only if users stored their keys with Microsoft.
• Risks:
  • Storing recovery keys on Microsoft’s servers enables both convenience and susceptibility to subpoenas.
  • Cited privacy experts warn this makes user data vulnerable, especially given requests by governments with poor human rights records.
• Comparisons:
  Apple and WhatsApp offer encrypted cloud key storage, but neither has reportedly surrendered user encryption keys.
• Memorable Quote:

  “Remote storage of decryption keys can be quite dangerous.”
  —Jennifer Granik, ACLU (12:52)

  • Microsoft averages around 20 requests for BitLocker keys per year; if keys are not stored with Microsoft, nothing is provided.

6. Deep Dive: Claudebot—Open Source Local AI Assistant (13:36–end)

What is Claudebot?

• Claudebot (spelled C-L-A-W-D-B-O-T):
  • An open-source project by Peter Steinberger.
  • Local LLM-powered agent you install and run on your own device (Mac Mini recommended).
  • Integrates with messaging apps (iMessage, Telegram, WhatsApp), so you communicate with it where you already chat.
  • Stores user preferences and memories as editable folders and Markdown files on your machine.
  • All local except for LLM calls; highly customizable and “tinkerable.”

Capabilities & Customization

• Full Local Control:
  • Can execute shell commands, write/install scripts, create new “skills,” access local files, and integrate with other services (e.g., Spotify, Philips Hue, Sonos, Gmail).
  • Grows more personalized the longer you use it.
• Self-Improvement:
  • Users can directly ask Claudebot to add new functionality, and it will “give itself” new skills as requested.
• Community:
  • Thriving open-source ecosystem—skills and plugins from users and from Steinberger.
  • Many are now buying Mac Minis just to run Claudebot instances.

Examples & Use Cases

• Federico Viticci shares he runs a version called “Navi”:
  • Knows his daily routines and preferences.
  • Communicates via Telegram; can handle audio messages, powered by ElevenLabs for TTS.
  • Once asked Claudebot to integrate with Google’s Nano Banana Pro image generation: it did, and handled API key storage securely.
  • Created a profile picture combining Navi’s crustacean mascot with the “Hey! Listen!” Zelda fairy—instructed Claudebot in natural language, which found and incorporated the meme on its own.
  • Adapted shortcuts for audio transcription using Whisper, all via natural chat commands.
• Quote:

  “Claudebot has fundamentally altered my perspective of what it means to have an intelligent personal AI assistant in 2026—would be an understatement.” (Federico Viticci, cited ~14:52)

• Host’s Take:

  “It’s an AI nerd's dream come true, and it’s a lot to wrap your head around at first.” (Brian, summarizing Viticci, 16:00)

Memorable Moments & Quotes (with Timestamps)

• “[TikTok] may now further leverage that info to serve you more relevant ads wherever you go online.” (TikTok privacy policy, 03:47)
• “[Apple] doesn't want to give potential bad actors more information than necessary.” (Brian, 05:59)
• “With this investigation, we will determine whether X has met its legal obligations under the DSA or whether it treated rights of European citizens… as collateral damage of its service.” (FT quoting EU tech chief Hanna Verkunen, 06:40)
• “The fact that LLMs cite sources such as Grokipedia… may in turn improve these sources’ credibility in the eyes of readers.” (Nina Jankowicz, 08:59)
• “Claudebot is… a tinkerer's laboratory…not poised to overtake the popularity of consumer LLMs anytime soon. Still, Claudebot points to a fascinating future for digital assistants.” (Federico Viticci, 14:48)
• “You can just do things... if you want Claudebot to give itself a piece of functionality that it doesn’t have… you can just ask it to do so and it'll do it for you.” (Federico Viticci, 15:44)
• “There’s been a run on Mac Minis of late specifically to use them as Claudebot boxes.” (Brian, 16:20)

Timestamps for Key Segments

| Segment | Time | |--------------------------------------|-----------| | TikTok privacy policy changes | 00:35–04:25 | | Apple AirTag second gen | 04:25–06:00 | | EU investigates XAI over Grok | 06:01–07:08 | | ChatGPT and Grokipedia | 07:09–09:34 | | Microsoft BitLocker keys | 11:50–13:35 | | Claudebot overview/deep-dive | 13:36–end |

Summary

This episode swiftly covers the biggest stories reshaping the tech landscape: user privacy shake-ups, innovative hardware upgrades, regulatory scrutiny of AI, and rising awareness of digital security risks. The highlight is a detailed look at Claudebot, an open-source local AI assistant that blends powerful LLM capabilities with user control, personalization, and the fun of open-source tinkering. Claudebot’s rise signals a grassroots movement towards self-hosted, steerable AI—embraced by power users and digital creators seeking to push the boundaries of what personal agents can do.

For those eager to explore the future of AI assistants beyond Big Tech silos, this episode offers an inspiring glimpse into what’s possible right now.