← Back to Model Beat
Products·Oct 5·all news from October 5, 2026

Our approach to EU text provenance rules

OpenAI is implementing invisible watermarking for text generated by ChatGPT in the European Union to comply with regional transparency regulations. While the tool demonstrates high detection rates under ideal conditions, performance declines significantly if the output is modified. The company plans to restrict initial access to researchers to evaluate the technology's effectiveness. Unlike some competitors, OpenAI will allow its API customers globally to opt out of these watermarking requirements, highlighting a flexible approach to implementing provenance standards across its developer ecosystem.

Covered by 6 sources

Related stories

ProductsBuilding advertising for the way people use AIOct 5 · 4 sourcesProductsIntroducing Playground: Create and play custom gamesOct 7 · 5 sourcesProductsWikimedia’s OpenAI Rebuke Points to Costs of AI Data ScrapingOct 5 · 7 sourcesProductsAI-powered hacking tools enabled a likely single attacker to breach multiple South Korean banksOct 6 · 7 sources