First off, keep in mind the join limits being decreased down to 100 https://discuss.dev.twitch.com/t/giving-broadcasters-control-concurrent-join-limits-for-irc-and-eventsub/54997. So legal aside for a moment, you’d be limited to 100 channels (and likely violating the Twitch Developer Agreement and/or ToS if you try to circumvent them) so you’d be very limited in scale.
Secondly, even if you anonymize the usernames, the content of the message itself may contain PII. Or if a person wishes to have their data removed from your service (as you do not own the rights to the chat message content) that would be very difficult if you’re scraping data without a channels/users consent.
Thirdly, this could fall under “Research - educational and non-commercial projects that examine trends over specific time periods” of the Developer Agreement, something which states Twitch will consider authorizing you to use the Twitch APIs to engage in the following activities only after entering into a separate agreement and running through an approval process:. So you’d likely require reaching out to developers@twitch.tv or legal@twitch.tv to get their explicit approval and agreement.
Finally, you should consult with your own legal expert. We’re not lawyers on this forum, and it’s likely none of the Twitch staff on this forum are from the legal department who would dispense legal advice here.