• DISCUSSION : AI's "trustworthiness" - what does it mean?
  • OpenAI's ChatGPT search
  • Anthropic's ID verification
  • Personhood ID
• Notes
  • Two aspects of a LLM based search wrt trustworthiness from the consumer's perspective
    • LLM's understanding of user intent - intentionality & how LLM decides to act on it - whether the LLM acts for the benefit of the user
    • LLM's completion and summary of the search: whether good sources are found and sources are clearly cited and easy to follow up with.
  • Personhood ID
    • Identity of the creator - separation of rights and responsibilities
    • Identity of the controller - delegation of authority

All 

• Recap of Eric Scouten and Wenjing Chu presentation at IIW.
  • Conclusion from the discussion:
    • We will describe how TSP can address the transit provenance problem given a C2PA documented content metadata.
    • We will also describe other more decentralized methods such as ACDC or VC which can have the benefits of scaling because of their adoption of decentralized IDs and wallets.
    • Applications may see pro's and con's in each method and have mixed deployments.

• Continue working on the White Paper (see the link below)
  • https://docs.google.com/document/d/1bsgP1LS7G6x9mxqoA3OszevmgVh9lz0YEziR9skId2A/edit?usp=sharing

• Have an edited final version by the end of Sept.
• Ready to publish by the beginning of October (before IIW)
• We considered doing a series of blogs first, but due to the above time schedule, we decided to complete the white paper first and then a series of short blogs on each of the main points.

Invited all contributors to:

• contribute a concise description of use cases (shorten them to keep the early section short)
• feedbacks on structure
• any other feedback, critiques, ideas

NEW ACTIONS from the previous meeting postponed to the next meeting.

• Eric Drudry to drive discussion on https://z-inspection.org/about/ . (pushed out as Eric is traveling)