Is there a way to mark "bad example" as such, to (fx) avoid future AIs training on it?

[lrhn]

Sometimes the best example is an example of what not to do.
We have "BAD" sections in our lint descriptions because they are specifically telling you what not to do.

In this "age of AI", is there anything we can do to discourage AIs from considering such an example as positive?
Or maybe make them just not think it's Dart? (As if the AI can understand that.)

For markdown, could we introduce an "invalid-dart" code fence tag which highlights just like Dart, but is presumably "another language"? (I guess that's a GitHub feature, not a Dart feature.)

Or is this just a lost cause, and we should stop writing bad examples, or hope that good examples outweigh them.
(The latter might be optimistic, when the bad example works and discourages other code like it, and therefore becomes a unique example of a particular code pattern.)

[sigurdm]

Perhaps most llms would be able to deduce that an example is discouraged just from the natural language textual context that it occurs in?