OpenAI AI Updates: October 6, 2026

1. OpenAI Will Watermark ChatGPT and Codex Text in the EU, Opt-In for API Users

OpenAI. OpenAI will embed an invisible statistical watermark called textGrain in ChatGPT and Codex text for EU users “over the coming weeks” to meet the EU AI Act’s machine-readable labeling requirement. API watermarking is opt-in worldwide, including through cloud partners like Microsoft Azure, which differs from Anthropic’s approach of watermarking Claude output globally. According to The Decoder, detection reaches about 95 percent on 400-token prose passages but only about 60 percent on math, and replacing 25 percent of the words pushes it below 20 percent. The detector is limited to selected researchers and organizations working under the EU’s Code of Practice on AI-Generated Content. Source (The Decoder, TechCrunch)

2. ChatGPT Adds Visual Ads Next to Image Generation Results

OpenAI. OpenAI introduced a visual display ad format that appears alongside image generation results in ChatGPT, starting later in October with a test group of advertisers in the U.S. and aimed at free and low-cost tier users. The company also added click attribution integrations with AppsFlyer, Adjust, Branch, Singular, Kochava, Northbeam, Triple Whale, and others, plus incrementality and geo-experiment partners such as Haus and Measured. For brand safety, OpenAI is running suitability pilots with DoubleVerify and Integral Ad Science. Source (TechCrunch)