Trains and answers
meta-externalagent: what it is and how to allow or block it
The same name decides two things: whether they train on your text and whether they use you to answer. Blocking it closes both doors at once.
What it is
Meta AI. Collects text to train Meta's models and for Meta AI's answers.
Like most AI crawlers, it reads the HTML exactly as your server sends it: whatever your site paints with JavaScript may never reach it.
How it identifies itself
meta-externalagent/1.1 (+https://developers.facebook.com/docs/sharing/webmasters/crawler)
This is what it sends in the User-Agent header. In robots.txt you only use the short name: meta-externalagent.
The lines for your robots.txt
Allow it
User-agent: meta-externalagent Allow: /
Block it
User-agent: meta-externalagent Disallow: /