OpenAI details new text watermarking system for ChatGPT, Codex, and the API

Billy Odell Tucker-Robinson Founder & Host — Banking With Billy Network • Intelligence Network • Data Science • AI Research • World News
Published: 2026-10-05T20:54:50.149Z • Permanent link

OpenAI, a leading provider of generative AI systems, has announced a significant development in its efforts to comply with the EU AI Act. The Act, which came into force in July 2023, requires providers of generative AI systems to mark AI-generated text in a way that makes it identifiable. To achieve this, OpenAI has developed a new text watermarking system that will be integrated into its ChatGPT, Codex, and API products.

According to a statement released by OpenAI, the new watermarking system will be based on a unique identifier that will be embedded in the text generated by its models. This identifier will be designed to be invisible to users but will be detectable by regulatory authorities and other third parties. The system will be rolled out in phases, with ChatGPT and Codex being the first to receive the new watermarking. The API will be updated later, with a specific timeline to be announced.

The development is seen as a significant step by OpenAI to demonstrate its commitment to transparency and compliance with regulatory requirements. The company has been under scrutiny in recent months over concerns about the accuracy and reliability of its AI models. By implementing the new watermarking system, OpenAI is seeking to build trust with its users and regulatory authorities, and to avoid any potential penalties or fines that could result from non-compliance.

The introduction of the new watermarking system has significant implications for the global infrastructure sector, particularly for companies that rely on OpenAI's products and services. For example, financial institutions and other organizations that use OpenAI's API to generate text-based reports and documents will need to ensure that the watermarking system is integrated into their systems. Failure to do so could result in significant reputational damage and financial losses.

Moreover, the new watermarking system is also expected to have a significant impact on the research community, which has been critical of OpenAI's lack of transparency about its AI models. Researchers who have used OpenAI's products and services in the past may need to re-evaluate their approach and consider alternative providers that offer more transparent and compliant solutions. The development is also likely to influence the development of new AI models and products, as companies seek to ensure that their systems are compliant with regulatory requirements.

The introduction of the new watermarking system is part of a larger trend towards increased regulation and transparency in the AI sector. The EU AI Act is just one example of the growing regulatory environment, which is expected to continue to tighten in the coming years. Other countries, such as the US and China, are also introducing their own regulatory frameworks for AI, which are likely to have a significant impact on the industry.