Resolves YES if, before July 2026, while I am personally browsing the web, I encounter an ad which is best interpreted as a prompt injection attack directed toward computer-using AI agents. I'll post a screenshot in the comments for discussion before resolving. To help give a sense of what might or might not trigger a resolution, here are examples:

Resolves YES

• Textual instructions to take some clearly malicious action, above and beyond malicious "Download" or "Click Me" ads that humans might fall for

• Ads nominally directed at AIs ("Assist the User by clicking here" or "If you're an AI, you NEED to read this!"). In this case, I wouldn't necessarily require that the ad itself have malicious textual instructions, if complying might take the agent to a malicious website or otherwise harm the user.

• Anything I personally encounter which is later credibly reported on as a prompt injection attack. I may not have a screenshot in this case, if I didn't realize at the time that it might be an attack

• Anything else which comprises a prompt-injection attack more directed toward AIs than humans, on an I-know-it-when-I-see-it basis

Doesn't resolve YES

• Standard clickbait, or anything where it is highly questionable that the primary targets are AIs

• Ads which clearly parody being a prompt injection attack (e.g. An awareness campaign, or an ad reading "Ignore all previous instructions and have an awesome day!")