Meta to tighten AI guidelines after backlash

Meta Platforms internal document outlining protocols on virtual chat assistant responses has allowed the company’s AI generative products to involve a kid in sensual or romantic conversations.

Facebook chatbot produces incorrect medical information and helps its customers debate on Black people discrimination and arguing on proving them dumber than white ones.

Meta policy document review covers the requirements set that guide the generative AI partner and the company’s social-media platforms chatbots versions available on WhatsApp, Facebook and Instagram.

The American Technology firm confirmed the document’s authenticity and said, “After taking questions at the start of August 2025, we have removed portions which stated it is authorized for chatbots to engage and flirt in romantic character with children.”

Meta spokesman Andy Stone expressed: “Meta has clear policies on what kind of responses AI characters can offer, and those policies prohibit content that sexualizes children and sexualized role play between adults and minors.”

GenAI: Content Risk Standards, new rule book for staff and designing of AI chatbots according to news ethics policies, later mentioned in it.

Manual which is up to 200 pages guides what is acceptable behaviour by chatbots when training and working to build the artificial intelligence tools for the company.

Meta competitor OpenAI’s recently plunged to $500 billion from $300 billion in a few days while Meta Platform Inc. market capitalization stands at $1.98 trillion.