OpenAI is preparing changes to how AI-generated content is identified and how advertising appears in ChatGPT. The company will introduce invisible watermarks for text generated by its AI models in the European Union, while also testing a new ad format linked to AI-generated images.
OpenAI brings invisible watermarking to AI-generated text
In a blog post on Monday, OpenAI said text generated by ChatGPT and Codex in the European Union will carry an invisible watermark over the coming weeks. API customers worldwide can also enable text watermarking for selected models, although the feature will remain disabled by default.
The move comes in response to the EU AI Act, which requires providers of generative AI systems to make machine-generated content identifiable. OpenAI has also started accepting applications from researchers and expert organisations to test its text watermark detector.
The system, called textGrain, adds an invisible statistical pattern to the model’s word choices. A detector then looks for the signal to determine whether a passage contains an OpenAI watermark. OpenAI said textGrain matched or exceeded other watermarking methods tested, including Google’s SynthID for text. The company also plans to make the underlying technology open source.
OpenAI noted that editing AI-generated text can weaken the watermark. In testing, the detector identified around 80 percent of watermarked 200-token passages at a 1 percent target false-positive rate. Detection increased to roughly 95 percent for 400-token passages in areas such as psychology, while results were less consistent in subjects such as mathematics.
For 400-token passages, replacing 10 percent of words with synonyms reduced detection from around 92 percent to 66 percent. Replacing 25 percent of words lowered detection to 17 percent.
The watermark cannot determine how much human input was involved or establish content ownership. It can indicate OpenAI involvement but should not be treated as proof of authorship.
ChatGPT to test new ad format
OpenAI will also begin testing a “new visual ad format” in ChatGPT later this month. The test will initially run in the US with a group of advertisers and will begin during image generation.
The ads will be labelled and kept separate from the generated image. They may show details about new products, how products are used or the experiences they enable.
OpenAI said the format will give advertisers more ways to reach customers. “As with our other ad formats, advertising does not influence the answers ChatGPT provides”, the company said.
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream LinkedIn | The Mainstream Facebook | The Mainstream Youtube | The Mainstream Twitter
About us:
The Mainstream is a premier platform delivering the latest updates and informed perspectives across the technology business and cyber landscape. Built on research-driven, thought leadership and original intellectual property, The Mainstream also curates summits & conferences that convene decision makers to explore how technology reshapes industries and leadership. With a growing presence in India and globally across the Middle East, Africa, ASEAN, the USA, the UK and Australia, The Mainstream carries a vision to bring the latest happenings and insights to 8.2 billion people and to place technology at the centre of conversation for leaders navigating the future.