Errata
TIP Protocol Whitepaper, Version 1.0
Current status
This page lists confirmed corrections and clarifications to the TIP Protocol Whitepaper, Version 1.0, in chronological order, each with the date of confirmation, the section affected, and the text as corrected or clarified.
2026-07-10 | Clarification | Section 7.2.1 Light Node
The first printing did not state a data-retention window for the Light Node. The specification is clarified as follows: the Light Node retains its entry-header and cryptographic-commitment data on a rolling one-year rotation, holding each entry for twelve months from the date it is observed and pruning it as newer entries are written, so the node continuously maintains a trailing twelve-month window of verifiable DAG history.
2026-07-12 | Clarification | Section 5.8 Content Classification and Raw Media Retention
The first printing did not describe how submitted media is handled during origin classification. The specification is clarified as follows: raw media files submitted for classification are held only for the duration of processing and in no event longer than twenty-one (21) days, after which they are permanently and completely deleted. Because a single submission may contain multiple files, media is processed through a queue. Only the classification output (the AI-generated probability and the declared Origin Code) and the content fingerprint are retained; the raw file is never published, sold, or written to the DAG. The origin classifier is improved over time by training on classification signals and provenance-dispute outcomes, not on the raw files, which have already been deleted. This clarification is reflected in the new Section 5.8.
How to report a suspected error
If you believe you have identified an error in the whitepaper, email [email protected] with:
- The section number and page (e.g., “ES.4, page 31”).
- The original text you believe is in error.
- The correction you propose.
- A reference (canonical source, regulation, spec, RFC) that supports the correction.
Reports are reviewed by The AI Lab and, where the error is confirmed, the correction is published on this page and reflected in the next printing of the whitepaper.
Document under review
For the current version of the whitepaper, see theailab.org/whitepaper.