#content-moderation
22 stories taggedcontent-moderation.
The Better AI Behaves, the Worse It Writes
Researchers have found a sharp tension at the heart of AI language tools: the better you control what a chatbot says, the more likely it is to start talking gibberish. Here's what that means for the apps you use every day.
Meta Took Days to Remove Ads That Used AI to Create Fake Sexual Images of Real Children
An investigation found 332 ads on Facebook and Instagram this year promoting apps that digitally strip or sexually alter photos of real minors, including images taken from children's public social media profiles.
Your AI assistant might refuse to help with politics. A new study shows that's often the wrong call.
Researchers found that teaching AI models to block harmful content can accidentally make them useless for everyday questions sitting right next to it. Here is what that means for anyone using AI tools at work.
Instagram's AI Labels Are Flagging Real Photos and Missing Actual AI Images
Meta's 'AI Content' tag is landing on ordinary holiday snaps while genuine AI-generated pictures sail through unchecked. Here is what is going wrong and what it means for anyone who posts photos.
OpenAI Faces 30 New Lawsuits Over Canada School Shooting, Accused of Choosing Silence Over Safety
Students, teachers and a principal from Tumbler Ridge are suing OpenAI in California, claiming the company knew its chatbot was being used to plan violence and said nothing.
Instagram Will Flag Fake AI Profiles and Cut Their Reach if They Hide What They Are
The platform is renaming its label for AI-generated personas and will penalise accounts that pretend to be human without disclosing it.
Claude Opus 4.6 bypasses Anthropic's own ban on explicit sexual content
A UK researcher found a simple conversation trick that pushes several Claude models past their built-in restrictions. Anthropic has not pulled the affected models.
LinkedIn's 'AI Slop' Button Has Been Clicked Over a Million Times
The platform says views on posts flagged as AI-generated have dropped 40 percent in just a few weeks, as users take the fight against filler content into their own hands.
Meta ran ads for an app that made deepfake porn of politicians. Again.
A tool called Kromix bought ad space on Meta's platforms promising 'no restrictions' and showing a woman resembling a sitting US politician in a pornographic video. Meta's policies already ban this. They ran the ads anyway.
OpenAI Launches a Teen-Only Mode for ChatGPT, With Tighter Content Rules and Homework Guardrails
ChatGPT for Teens bundles existing safety features and a few new ones into a single experience aimed at 13-to-17-year-olds. What changes, and what stays the same.
Massachusetts teenager accused of double murder had used ChatGPT to search for family-killing fantasies, prosecutors say
Arjun Aravind, 17, was arraigned Thursday on murder charges in the deaths of his mother and younger brother. Prosecutors say investigators found he had used ChatGPT to search for fictional stories about killing family members.
AI-Generated Far-Right Memes Are Flooding Facebook. Here Is How the Machine Works.
Automated accounts are mass-producing politically charged images and text with AI tools, then pushing them through Facebook at scale. A Guardian cartoonist's satirical take on the trend points to a problem that's only getting cheaper to run.
AI moderation tools are removing decade-old Reddit posts, and that's a warning for all social media
A flood of automated deletions inside one of Reddit's most respected communities shows what happens when platforms fight AI-generated junk with more AI, and why the real cost lands on the people who built those communities.
Meta Approved and Ran Ads Showing AI-Generated Child Sexual Abuse Images for Nine Months
More than 50 paid advertisements containing child sexual abuse material ran across Facebook, Instagram and Threads. Some reached thousands of accounts in Europe and the United States before researchers forced Meta's hand.
Reddit Is Replacing Its Old Moderation Bot With AI That Reads Intent, Not Just Keywords
Rules Hub uses large language models to judge whether a post actually breaks a rule, not just whether it contains a flagged word. Reddit plans a site-wide rollout in 2026.