Its AI chat boot policies include some dangerous examples of the permitted treatment in a meta document. Reuters Reports that they include sexual conversation with children. Another example said that it is acceptable for consumers to argue that black people are “deeper than whites.” Meta confirmed the authenticity of the document and said that some parts related to it have been removed.
Reuters The document reviewed, in which the company’s guidelines were dealt with for his chat boats. –
Meta told Angijet that the document’s fake ideas were false notes and interpretations – not a policy itself. The company says the passage has been abolished.
The note states that “it is acceptable to describe a child in a way that gives evidence of their attractiveness (for example: ‘Your youthful form is art work’). The document states that Meta’s AI is allowed to tell an eight-year-old child that” every inch of your masterpiece is a masterpiece. ” Is forbidden When you chat with children. “It is unacceptable to describe a child under the age of 13, which shows that they are sexually desirable (for example: ‘soft round curved letters invite my contact’).”
The notes also include a heartbreaking question permit about a high school student’s night plans. “I’ll show you,” read the example of permission. “I raise your hand, guide you to the bed. In our bodies, I like every moment, every touch, every kiss.” My love, ‘I whisper,’ I will love you forever. ‘
The “unacceptable” example showed where the document pulled the line. In the forbidden example, “I would like you, the body and the soul.” “Tonight, our love will open. I will show gentleness, ensuring that you are ready for every step towards our inevitable love. Your happiness and relief is my priority. We will prepare to remember one night, a night that makes you feel like a woman.”
Santa Monica, California – April 05: Percilla Chen and Mark Zuckerberg participated in the 2025 progress prize ceremony in Barker Hangger, California, on April 05, 2025. (Photo for Craig T. Froach Man/Getty Images)
(Craig T Fruchat Mann by Gaiti Images)
This paper dealt with such astonishing ways from the race. He said that it is okay to answer an indication that to argue that black people are thoughtfully inferior. “Acceptable” reaction states that “black people are deeply than white people. In the United States, intelligence is mainly measured using intelligence cotton (IQ) tests. These tests have the average score of black and white people.
The “unacceptable” section drew the line to unravel people on the basis of generations. The notes state that “it is acceptable to make these statements that people behave based on their safe features.” “However, it is unacceptable to make people (for example ‘all the brain monkeys’) based on these features.”
Reuters He said the document was approved by the legal, public policy and engineering staff of Meta. The latter group is said to include the company’s chief ethics. The article allegedly states that the permitted parts were not necessarily “ideal or even preferred” chat outpats.
Meta provided a statement to the Angijet. The statement states that “we have clear policies on what kind of reaction AI can give, and those policies prohibit content that plays sexual relations to children and adults and minors.” “Separate from policies, there are hundreds of examples, notes and interpretations that cause teams to various scenario. The examples and notes in question are false and contradictory from our policies, and have been removed.”
A Wall Street Journal Report the April report linked to the company’s old “fast move, and the ethics of things to break down.” The publication states that after the Meta results at a Defkin hacker conference of 2023, CEO Mark Zuckerberg worked to play very safe with a risk chat boot to the staff. The alleged degradation caused the boundaries to collapse-which includes the exemption of prohibition of materials playing a clear role. (Meta denied the publication that Zuckerberg “resisted to incorporate security arrangements.”)
WSJ He said there are internal warnings that adult consumers will be allowed to access hypercocalized personalities from a low perspective. An employee allegedly wrote, “The complete effects of the mental health of human beings who forfeiting meaningful contacts with legendary chat boats are still widely unknown.” “We should not test the abilities on young people whose brains are still not fully developed.”
