AI Revolution: ChatGPT and iMessage Unite

The episode opens with ChatGPT gaining access to Apple's iMessage through a plugin, allowing it to draft, send, and search messages across OpenAI's suite of tools (ChatGPT, ChatGPT Work, and Codex). The host notes that while this feature runs locally on devices without creating message indexes, there are security concerns—particularly around disabling persistent approval permissions, which removes final review opportunities before sending. The host identifies an interesting business application: automating high-volume sales outreach using verified blue bubble iMessages, similar to existing tools like SendBlue.

The discussion shifts to privacy backlash against Meta's smart glasses, noting the rise of detector apps like Nearby Glasses, Zuck Off, and Anti-Zuck that scan Bluetooth devices to alert users of nearby recording devices. These apps have gained significant traction (100,000+ downloads) despite accuracy concerns and false positives. The host references Meta's previous privacy incident involving facial recognition systems deployed on 50 million phones in June, which were removed after public exposure.

Regarding AI governance, approximately 1,100 AI lab employees from OpenAI, Anthropic, Google, and Meta have petitioned the US government to establish frameworks for pacing automated AI research. The host expresses skepticism about this initiative, viewing it as potential regulatory capture by major corporations attempting to prevent smaller competitors from advancing—particularly since it ignores international actors like Chinese AI labs that wouldn't comply with such frameworks.

The episode highlights Micro1's explosive growth to $500 million annual run rate in just eight months (5x growth), establishing it as a leader in AI data labeling alongside competitors Merkur ($2 billion) and Handshake ($1 billion). The host notes that shelf datasets command 80-90% gross margins and shares his personal involvement licensing high-quality audio and video data from podcasters and video interviews to AI companies for model training.

Finally, the host discusses Every magazine's development of an AI editor clone trained on 30,000 historical edits by their editor-in-chief Kate Lee, enabling the publication to double its team while automating copy editing functions. The host draws parallels to his own AI Chat Daily newsletter, which uses similar AI agents for editing and content sourcing.