RE: LeoThread 2025-02-22 00:52
You are viewing a single comment's thread:
This issue predominantly affects Europe rather than being solely an AI-related concern.
"In many ways, the dangers posed by AI feel abstract to most, but for some, they are very tangible."
0
0
0.000
In just two weeks, there’s the possibility of facing years behind bars due to the government's utilization of an AI tool in a group chat labeled “shitposting.”
This AI system assigned a 'toxicity score' to every message and deemed most as harmful. One of the messages attributed to me was: "Some cultures are inferior, backwards cultures that still practice Female Genital Mutilation, for example."
This message received a perfect score of 100 for toxicity, sexism, and racism. It's noteworthy that this was classified as sexist by the AI—something punishable under Belgian
law.
Most messages labeled as 'toxic' by the AI were misinterpreted. For instance, any mention of "an African" was marked as racist, and phrases like "leftist tears" received the highest toxicity level.
I (allegedly) quoted a black journalist who made a racist statement, but the AI misunderstood and wrongly attributed the quote to me, contributing to the total toxicity score that has led to a one-year prison sentence and a €16,000 fine.
Defending against this is virtually impossible, as the Public Prosecutor relies on the 'Total Toxicity Score' for evidence, rather than examining the quoted content in question.
The Public Prosecutor’s interpretation of 'shitposts' is also ridiculously broad: “Shitposts are intentionally insulting messages meant to provoke.”
An appeal is scheduled for next week.