How should we deal with questions that have a high probability of an unethical use case?

Question

I am referring to this question. The author asks if there exists a possibility to create deepfake-videos on a low-end PC.

On the one hand, I want to assume the best in people and therefore do not want to accuse them of an unethical use case. However, on the other hand, I also don't want to be naive: the majority of the videos created with deepfake are unethical.

Also, even if the author has good intentions and is planning to use deepfake in an ethical way, other people who stumble upon this question may not. This can, of course, happen with any question on Stack Exchange. The difference, however, is that the probability is very high that the answers to this question will be used with unethical intentions.

Further, there are also questions like this one where the author is interested in the functionality of deepfake but is at the same time using it to create videos. Deepfake is indeed a very fascinating piece of software and explaining it could add valuable knowledge to this site. But the probability is still high that by answering it, we support unethical behavior.

To put it in more general terms: how should we deal with questions that have a high probability of being used unethically by the author or by other people? Deepfake is most probably just the beginning.

Edit: I'd like to add a short description of the status quo to underline the importance of this question. Currently, answers to questions regarding deepfake are encouraged by the site mechanics. If one posts a valid answer, he or she gets rewarded with 10 points per upvote. That means, if we do not agree upon how we should deal with such questions, we passively support them.

Answer 1

Since this question is unanswered for a week now, I would like to contribute a suggestion.

Questions that are on-topic on the AI SE include the theory/concepts of AI, social issues, and so on. Questions regarding the implementation of algorithms are definitely off-topic.

I suggest the community should be allowed to discuss the theory and the theoretical concepts of algorithms that have a high likelihood of being used in an unethical way as long as the discourse stays purely theoretical. Questions regarding the implementation, or the application of such algorithms should be flagged as off-topic (what they actually are) and closed. These questions should be treated more strictly (when in doubt, close them) since they are not only probably off-topic but also have a higher probability of being used with unethical intentions.

That means, a question like How to make deepfake video without a fancy PC? should be closed and a question like What are the differences between Deepfakes, FaceSwap and Face2Face? should be left open.

An open issue is still where to draw the line. I think this has to be evaluated on a case by case basis. In the end, questions about the implementation of an algorithm are off-topic anyways.