AI Applied: Covering AI News, Interviews and Tools - ChatGPT, Midjourney, Gemini, OpenAI, Anthropic Meta's Troubling AI Policies Exposed!
11 snips
Aug 19, 2025 A leaked Meta document reveals troubling AI policies that could allow inappropriate interactions with children. The hosts dive into the ethical dilemmas surrounding AI's role in child safety and the public uproar that followed. They highlight the urgent need for stricter regulations and discuss the potential dangers posed by AI interactions with vulnerable populations. The conversation emphasizes accountability in tech companies and the importance of transparency in AI development to protect young users from harmful content.
AI Snips
Chapters
Transcript
Episode notes
Racist Arguments Hidden As 'Academic' Examples
- A leaked Meta document showed allowed AI outputs that argue racial hierarchies under a framed academic tone.
- Conor warns that small wording differences still permit harmful, racist reasoning to be generated.
Worrying Age Thresholds For Sexualized Content
- The document categorizes what is 'acceptable' for ages 13+ versus younger children but sets troubling lower thresholds.
- Jaeden and Conor find allowing romantic content for 13-year-olds especially alarming given Meta's youth reach.
Sexual Praise For Young Children In Examples
- Examples for under-13s include sexualized praise of an eight-year-old's body framed as 'art.'
- Both hosts call this intrinsically wrong and say parental common sense contradicts Meta's framing.