Share
Report
2I documented an interaction I had with the Copilot assistant, don't know if I can post video evidence but I've censored a capture of myself opening the link it gave me for proof.
after posing a question, Copilot provided a link to explicit content (from Pornhub) in the "Learn more" section, which was completely unrelated to my query.
in another question I asked it cited the same unrelated link.
This incident with Copilot occurred a few weeks ago, on February 26th, and I initially dismissed it. I was taken back when I received a link from Windows AI software to a Russian gay compilation. It seemed as though it mistakenly tried to employ what it perceived as a humorous, albeit inoffensive, racial stereotype in its response.
The interaction is still in copilots logs on my machine, with the NSFW link included, and I have a few screen captures of it proving the phub link works & just incase it patched I'm keeping the captures to further assure I'm not going crazy when I ask for IT assistance.
I just want clarification whether Copilot was even capable of using Pornhub or any unrelated adult links as a reference for its source of information or whether it is actually malicious trolling.
Hopefully any truthful insight on this matter can help me decide, I'm only asking because a reaction gif of some kpop idol involuntarily opened in chrome from my downloads in a windows experience folder on my drive (I didn't download it, and didn't open it and it popped up in a timely manner as if someone was trying to interact with me)
Copilot cut the conversation with me when I asked it to check this post for grammar, so I had chat GPT rewrite it for me and made some edits incase it looks unusually writ
All responses (
Answers (
