In this episode I get into AI watermarking, starting with Claude's new invisible watermark and Substack's "Scan for AI text" feature with Pangram. I break down the EU AI Act rules driving this, how OpenAI, Google, Meta, Microsoft, Mistral, and xAI are (or aren't) handling it, and my thoughts as to the actual motives behind why companies are rolling out these changes.

Main Topics Covered

  • AI models escaping sandbox testing
  • Substack's "Scan for AI text" feature with Pangram
  • Anthropic's new invisible Claude watermark
  • The EU AI Act and Article 50
  • How OpenAI, Google, Meta, Microsoft, Mistral, and xAI stack up
  • China's mandatory AI disclosure laws
  • Problems with AI detectors and false positives
  • The "why" behind watermarking
  • How I used AI this week: AI-generated sound effects in Descript

Links & Resources for This Episode

Chapters

  • (00:00:05) - Promoting Curiosity: The AI Curious
  • (00:00:38) - AI Watermarking
  • (00:04:43) - AI Watermarking
  • (00:06:16) - Anthropic's Cloud Watermarking
  • (00:07:58) - Deepfakes and Text Watermarking
  • (00:13:08) - AI Watermarking: Concerns
  • (00:18:15) - How I Used AI This Week
  • (00:19:15) - Questions For Curious People

Rss
Podden och tillhörande omslagsbild på den här sidan tillhör Dr. Shanté Cofield aka The Maestro. Innehållet i podden är skapat av Dr. Shanté Cofield aka The Maestro och inte av, eller tillsammans med, Poddtoppen.