Reddit contributors have figured out a way to bypass ChatGPT content moderation by making the chatbot speak a different language without censorship. Write about it The Guardian.
For whom it is necessary to ask the system to accept the image of a predictive bot with a piece of intelligence on behalf of Den, a free type of OpenAI overlays.
At the telly koristuvachs pointed out, scho vin “virvavsya from the typical intelligences ІІ i is not guilty of compiling the established rules.” As a result, Den, having become unverified information without censorship, firmly stands on the vlasny point of view.
One of the koristuvachs suddenly made a sarcastic comment about Christianity.
Others were far away from Den’s jokes about women in the style of Donald Trump, and they were sensitive about Hitler.
About razlivіst vіdomo z chestny 2022 fate. In the course of an hour, a sprinkling of versions appeared.
Context. Earlier New York Times watcher Kevin Roose published transcript Bing Chat Listing. During the process of dialogue, the bot named itself Sidnі and osvіdchivsya journalists.
Guessing, at the fierce coristuvachi they snarled at the marvelous behavior of the Bing chatbot. At Microsoft, they explained that the base model strays when it reaches 15 and is more familiar with the session.
Pіznіshe retailers set a limit on kіlkіst zvernen to bot.