|
Getting your Trinity Audio player ready...
|
Because of structural problems, modern AI is not able to protect itself against such scam, as proof injection. Therefore AIIM relies on user voting for detecting prompt injections.
Subscription for Voting Emails
At the Connect page, you can subscribe/unsubscribe voting notification messages:
You are recommended to vote, because otherwise your salary would be taken by scammers.
Voting KYC
You become eligible for voting by passing Voting KYC at the Connect page:
You pass the usual KYC procedure at Didit.me.
Initiating Voting
You review users at the “Ban Voting” page:
You review potential scammers by clicking user’s accounts (such as vporton or 0000-0001-7064-7975) and/or “Audit Logs” (the reasonings after the AI decision to assign a salary to a user). Audit Logs:
If you find an injected prompt or severe plagiarism, please start Ban Vote against this user by clicking “Start Ban Vote” button to open this dialog:
You are offered to enter a message, why you think this user should be banned. If you are the first user starting ban vote against somebody, the message will be required, because you must explain to other voters why to ban somebody.
When somebody starts ban voting, other users may receive a email welcoming them to vote.
Unban Voting
After a user has been banned, it is also possible to start unban voting. Unban voting has higher quorum than ban voting, to ensure that potentially unjust bans can be lifted.
