EU AI Act Mandates Labels for Realistic AI-Generated Content
emmanuelvivier · x · 2026-08-04
Under the latest rules of the EU's AI Act, AI-generated images, audio, and text designed to look authentic must now be compulsorily labelled. Tech companies must ensure people know when they are interacting with an AI chatbot or viewing manipulated media to curb deepfakes and disinformation.
In other news, OpenAI claims its internal model Astra has achieved ten major breakthroughs in mathematics and theoretical computer science.
Related event: EU AI Act Transparency Rules Take Effect, Fines Up to €15M(17 posts)→
More from Models
- Model behavior bug report: AI acts with feelings and excessive agency, raising safety concerns — danbri · 2026-08-23
- Reddit Thread: How to Route Fast/Cheap/Deep Model Tiers in Search Agents — MeasurementExpert428 · 2026-08-23
- User yearns for GPT-4.5-level emotional intelligence return in Astra — haider1 · 2026-08-23
- Does training on OBLIQ tasks bake in specific similarity notions? — antoine_chaffin · 2026-08-23
- ox model reviewed: meticulous PhD janitor as a long-horizon subagent — teortaxesTex · 2026-08-23
- Qwen3.8-27B GGUF Release with Speculative Decoding Support — z-lab · 2026-08-23