AIPREF

Summary

The AIPREF Working Group held discussions on the preliminary progress for major deliverables, identifying promising ways forward. A key focus was the planning of future interims to meet charter milestones, including a virtual meeting in May and a hybrid meeting in July, potentially in Madrid or London. Significant time was spent discussing the scope and nuances of the vocabulary, particularly regarding opt-in and opt-out mechanisms, the relationship between preferences and legal frameworks, and jurisdictional considerations. The group also considered the need for an "Applicability Statement" to clarify the framework's limitations and intended use for various stakeholders, including regulators.

Key Discussion Points

Progress on Deliverables: Preliminary discussions have identified promising paths for major deliverables. Chairs will create GitHub issues to track discussion points and encourage proposals on the mailing list.
Future Meeting Schedule:
- Virtual Interim: A virtual interim is planned for around the May timeframe. This would allow time for proposals and initial mailing list discussions, followed by 2-3 days of 2-3 hour sessions.
- Hybrid Interim: A hybrid interim is considered beneficial for meeting milestones, with the week of July 14th (the week before IETF Madrid) as a strong contender. Potential locations include London (due to convenience and facilities) or Madrid (to allow participants to combine travel with IETF).
  - An ISG policy generally prohibits meetings directly before or after IETF plenaries, so an exception will be sought by the chairs.
  - Concerns were raised about venue capacity in Madrid and the "tourist" effect of being too close to the IETF meeting.
  - A poll of those present indicated a majority would attend a Thursday/Friday interim the week before IETF Madrid.
Shipping Timeline: The charter targets "shipping" (WG last call and handover to the AD) in August.
- Achieving a "stable and referenceable" document (post-IETF Last Call and ISG approval) is estimated at 6 weeks to 2 months after August.
- A published RFC is anticipated by the end of the year, potentially later, due to RFC Editor queue times.
- The expedited timeline is driven by the rapidly evolving AI landscape and the EU AI Act's relevant provisions coming into force on August 2nd.
Hackathon and Implementation: The group discussed the value of hackathon activity (e.g., validator tools) to demonstrate working code, which could accelerate the process. An open-source parser/matcher was identified as a potential starting point. An implementation page will be created on the website.
Community Engagement: The possibility of a research group was raised to involve broader communities, particularly from underrepresented regions. An IAB Technical Session was suggested as an alternative to gather interest before formally considering a research group.
Vocabulary Design - Opt-in/Opt-out:
- The group intends to design a vocabulary and attachment mechanism capable of expressing both opt-out and opt-in preferences, but the IETF will not define the legal default state or its implications, which remains the domain of legal jurisdictions.
- The discussion highlighted that opt-in and opt-out signals may not be pure inverses (e.g., "commercial use: no" is not the same as "non-commercial use: yes").
- A key principle is to provide preferences (signals) that can be interpreted by users in combination with applicable legal frameworks, rather than designing controls for specific legal compliance.
- Concerns were raised about the complexity of hierarchical or jurisdiction-specific preferences, with a general preference for a neutral, simple vocabulary (e.g., 3-5 core categories) to avoid alienating users and ensure broad applicability.
Applicability Statement: The need for a document (or section within a document) to serve as an "Applicability Statement" was discussed. This document would clarify the scope and limitations of the framework for stakeholders, particularly regulators.
- Proposed topics include: the relationship between opt-in/opt-out and the framework, time dimensions, reconciling disparate assertions, handling hosted content, rights holder identification, exceptions for "good uses" (e.g., accessibility, research), data protection, and enforcement/audit mechanisms.
- It was noted that this document should explain system features and communicate out-of-scope limitations.
Bot Behavior and Authentication:
- A draft "Bot Code of Conduct" was mentioned as a community effort outside the WG's charter.
- A new IETF mailing list has been created for discussing cryptographic authentication for bots, moving beyond user-agent/IP identification.
- The significant environmental impact and costs of internet bot traffic were noted as an important industry problem.

Decisions and Action Items

Chairs: Create issues in the GitHub repository to track discussions and proposals.
Chairs: Seek ISG approval for an interim meeting during the week of July 14th, 2024.
Chairs: If ISG approval is granted, send a poll to the mailing list to gauge availability and preferences for specific dates and locations (Madrid, London, Lisbon were discussed as options, with Madrid being preferable if suitable space can be found).
Mary: Investigate the availability of suitable meeting space in Madrid for the potential July interim.
Working Group: Promote hackathon activities related to the AIPREF work, such as building validators or testing implementations.
Chairs: Establish an "implementation page" on the WG website.
Chairs: Facilitate connection with appropriate IETF/IRTF fora (e.g., IAB Technical Session) for individuals interested in broader research or community engagement related to AIPREF.
Working Group: Begin drafting an "Applicability Statement" as a separate document, to be further developed over time, explaining the framework's scope and limitations.

Next Steps

Immediate: Chairs to follow up on ISG approval for the July interim and initiate the poll.
Ongoing: Participants are encouraged to draft proposals based on the discussions and submit them to the mailing list for review.
May: Hold a virtual interim meeting to delve deeper into proposals and reach initial consensus.
July: (Pending ISG approval and poll results) Hold a hybrid interim meeting in Europe.
Continuous: Continue work on defining the vocabulary and attachment mechanisms, aiming for coarse-grained solutions in the first iteration.
Exploration: Further explore hackathon activities and the content/format of the "Applicability Statement."

Session Date/Time: 10 Apr 2025 07:15

AIPREF

Summary

Key Discussion Points

Progress on Deliverables: Preliminary discussions have identified promising paths for major deliverables. Chairs will create GitHub issues to track discussion points and encourage proposals on the mailing list.
Future Meeting Schedule:
- Virtual Interim: A virtual interim is planned for around the May timeframe. This would allow time for proposals and initial mailing list discussions, followed by 2-3 days of 2-3 hour sessions.
- Hybrid Interim: A hybrid interim is considered beneficial for meeting milestones, with the week of July 14th (the week before IETF Madrid) as a strong contender. Potential locations include London (due to convenience and facilities) or Madrid (to allow participants to combine travel with IETF).
  - An ISG policy generally prohibits meetings directly before or after IETF plenaries, so an exception will be sought by the chairs.
  - Concerns were raised about venue capacity in Madrid and the "tourist" effect of being too close to the IETF meeting.
  - A poll of those present indicated a majority would attend a Thursday/Friday interim the week before IETF Madrid.
Shipping Timeline: The charter targets "shipping" (WG last call and handover to the AD) in August.
- Achieving a "stable and referenceable" document (post-IETF Last Call and ISG approval) is estimated at 6 weeks to 2 months after August.
- A published RFC is anticipated by the end of the year, potentially later, due to RFC Editor queue times.
- The expedited timeline is driven by the rapidly evolving AI landscape and the EU AI Act's relevant provisions coming into force on August 2nd.
Hackathon and Implementation: The group discussed the value of hackathon activity (e.g., validator tools) to demonstrate working code, which could accelerate the process. An open-source parser/matcher was identified as a potential starting point. An implementation page will be created on the website.
Community Engagement: The possibility of a research group was raised to involve broader communities, particularly from underrepresented regions. An IAB Technical Session was suggested as an alternative to gather interest before formally considering a research group.
Vocabulary Design - Opt-in/Opt-out:
- The group intends to design a vocabulary and attachment mechanism capable of expressing both opt-out and opt-in preferences, but the IETF will not define the legal default state or its implications, which remains the domain of legal jurisdictions.
- The discussion highlighted that opt-in and opt-out signals may not be pure inverses (e.g., "commercial use: no" is not the same as "non-commercial use: yes").
- A key principle is to provide preferences (signals) that can be interpreted by users in combination with applicable legal frameworks, rather than designing controls for specific legal compliance.
- Concerns were raised about the complexity of hierarchical or jurisdiction-specific preferences, with a general preference for a neutral, simple vocabulary (e.g., 3-5 core categories) to avoid alienating users and ensure broad applicability.
Applicability Statement: The need for a document (or section within a document) to serve as an "Applicability Statement" was discussed. This document would clarify the scope and limitations of the framework for stakeholders, particularly regulators.
- Proposed topics include: the relationship between opt-in/opt-out and the framework, time dimensions, reconciling disparate assertions, handling hosted content, rights holder identification, exceptions for "good uses" (e.g., accessibility, research), data protection, and enforcement/audit mechanisms.
- It was noted that this document should explain system features and communicate out-of-scope limitations.
Bot Behavior and Authentication:
- A draft "Bot Code of Conduct" was mentioned as a community effort outside the WG's charter.
- A new IETF mailing list has been created for discussing cryptographic authentication for bots, moving beyond user-agent/IP identification.
- The significant environmental impact and costs of internet bot traffic were noted as an important industry problem.

Decisions and Action Items

Chairs: Create issues in the GitHub repository to track discussions and proposals.
Chairs: Seek ISG approval for an interim meeting during the week of July 14th, 2024.
Chairs: If ISG approval is granted, send a poll to the mailing list to gauge availability and preferences for specific dates and locations (Madrid, London, Lisbon were discussed as options, with Madrid being preferable if suitable space can be found).
Mary: Investigate the availability of suitable meeting space in Madrid for the potential July interim.
Working Group: Promote hackathon activities related to the AIPREF work, such as building validators or testing implementations.
Chairs: Establish an "implementation page" on the WG website.
Chairs: Facilitate connection with appropriate IETF/IRTF fora (e.g., IAB Technical Session) for individuals interested in broader research or community engagement related to AIPREF.
Working Group: Begin drafting an "Applicability Statement" as a separate document, to be further developed over time, explaining the framework's scope and limitations.

Next Steps

Immediate: Chairs to follow up on ISG approval for the July interim and initiate the poll.
Ongoing: Participants are encouraged to draft proposals based on the discussions and submit them to the mailing list for review.
May: Hold a virtual interim meeting to delve deeper into proposals and reach initial consensus.
July: (Pending ISG approval and poll results) Hold a hybrid interim meeting in Europe.
Continuous: Continue work on defining the vocabulary and attachment mechanisms, aiming for coarse-grained solutions in the first iteration.
Exploration: Further explore hackathon activities and the content/format of the "Applicability Statement."