potiuk commented on PR #2:
URL: https://github.com/apache/comdev/pull/2#issuecomment-4320194185

   @grobmeier — pinging you in your VP, Data Privacy capacity. There's an 
unresolved policy question on this PR (an MCP server that lets users query 
Apache mailing list archives via an LLM) where your guidance would help.
   
   The concern, raised by @sebbASF: private ASF lists may carry PII with the 
expectation of non-propagation; feeding their content to an LLM (especially a 
hosted one) plausibly violates that expectation.
   
   The proposed posture (in #3, on top of this branch):
   
   - **Block all private lists by default.** Two layers — pattern blocks for 
known-private list names, plus checking PonyMail's own `private: true` flag on 
every response so unconventionally-named private lists are also caught.
   - **Opt-in** via a `PONYMAIL_ALLOWED_LISTS` env var, so a contributor 
authorized to access (e.g.) their own project's `security@` can do so while 
taking responsibility for the LLM environment they use.
   - **Documentation** explains the hosted-vs-local LLM risk, links the ASF 
Mailing Lists / Privacy Policy pages, and flags this as an interim posture 
pending updated ASF guidance for LLM use specifically.
   
   Two questions where your read would help:
   
   1. Is "block all private lists by default + per-list opt-in by an authorized 
human operator" an acceptable interim posture while ASF privacy policy for LLM 
use is being clarified?
   2. Should certain categories of private list (e.g. `members@`, 
board/secretary/treasurer, `private@`) be policy-blocked even with opt-in, 
while operationally-private lists like `security@` could be operator's 
discretion?
   
   No urgency on the merge — happy to hold until you've had a chance to weigh 
in.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to