Time

Agenda Item

Lead

Notes

5 min

• Start recording

• Welcome & antitrust notice

• Introduction of new members

• Agenda review

• Assigning minutes taker

Chairs

• Antitrust Policy Notice: Attendees are reminded to adhere to the meeting agenda and not participate in activities prohibited under antitrust and competition laws.

• ToIP Policy: Only members of ToIP who have signed the necessary agreements are permitted to participate in this activity beyond an observer role.

• ToIP TSWG IPR Policy: see TF wiki page. AI & Metaverse Technology Task Force. Please remember that TSWG2 has transitioned to a new IPR policy using OWF license.

15 mins

• Introduction of new members

• Any general announcements news, that could be of interest to the TF

• Last week's action items

All

•  

  • Discussions on how to promote ToIP in LF projects

  • Presentations by @Judith Fleenor and @Wenjing Chu

  • Trustworthiness of open-source software & Cyber Resilience Act will be required by 2027

  • Workshop:  Open Source Software Stewards and Manufacturers Workshop | LF Events

  • Reputation vs Identity of the Committers & Maintainers 

  • Will be CRA "clearing houses" 

  • Pushback by contributors stating that it's open source and should be anonymous, that's why Reputation is more acceptable.

  • @Wenjing Chu will be presenting "From Security to Trust" in the afternoonLFMS news (LF Member Summit is Nov 19-21 in Napa, CA. 25 years of LF.

• Holiday calendar:

  • 05 Dec is on

  • 19 Dec (@Wenjing Chu out) is off

  • first meeting in 2025 is off

15 mins

White Paper & C2PA and CAWG, TSP

All 

• Current version

• Create a new version from scratch in the new year

• Publish date: to be discussed

10 mins

@Eric Drury AI Safety Businesses & Tools

All

Eric moderated a panel on Cooperative AI Governance' as part of the Paris Peace Forum, and met with a couple of guys who run AI safety businesses with tools "designed to probe the robustness of LLMs against prompt injections" for example and will discuss these kinds of products / businesses - red teaming, blue teaming,

• All Tech is Human

• All about governance

• Macron wanted global collective functioning governance

• Company on panel on AI compliance: https://alltechishuman.org/ & https://www.prism-eval.ai/

• @Steven Milstein Reputation and data provenance are more important than identity. 

• @Neil Thomson Consumer Reports at IIW may be getting into the business of reputation

• @Savita Farooqui Accountability in Government space is more important than bot or human

• @Wenjing Chu Regulations for Accountability 

• Example: Self-driving cars. Who's accountable? Driver not manufacturer.

• @Eric Drury: Red team: self-auditing vs Blue Team: external 

• @Wenjing Chu: Provenance of AI data set

  • Tools like C2PA & Trust Spanning Protocol 

  • Boston group of doctors: One group with ChatGPT trained for doctors, one with ChaptGPT not trained, one no ChatGPT.

    • Result: [ ] @Wenjing Chu please complete 

      • Generic ChatGPT had the greatest success rate >90%

      • Use of LLM by the doctors does not improve diagnosis success rate of ~75%

      • Suggested reasons: doctors are not trained to effectively use LLMs to improve diagnosis, they do not trust LLMs.

    • See reference: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2825395

10 mins

Discussions of guardrails for AI - the lack of, or the insufficient of

5 mins

Chairs

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 travelling)