Generative AI watermarking technologies embed invisible markers in AI-generated content – including text, images, audio and video – to verify authenticity and help trace content origins. As AI-generated content becomes increasingly hard to differentiate from that created without AI, there has been a surge in innovative watermarking technologies147 designed to help combat misinformation, protect intellectual property, counter academic dishonesty and promote trust in digital content. Watermarking techniques aim to subtly alter generative AI outputs without noticeably impacting their quality. Text-based watermark technologies, such as Google DeepMind’s SynthID technology,148 take advantage of the fact that there are thousands of words in a given language that can be randomly substituted by others. They work by including a narrow and specific subset of such words throughout AI-generated text that seems natural but is distinct from the more random word choices a human writer might make. This results in an AI-specific textual “fingerprint”. Image and video watermark technologies include introducing imperceptible changes at the pixel level that can survive edits like resizing and compression – for instance by subtly altering the values of individual pixels so that a machine can see the changes, but the human eye cannot, or embedding hidden patterns149 in generated output that only a machine can extract. Watermarking AI-generated content gained traction in 2022, as models like ChatGPT and Stable Diffusion gained popularity and widespread use. By 2023, major AI companies, including OpenAI, Google and Meta, committed to watermarking under regulatory pressure.150 A breakthrough came in 2024 when Google DeepMind open-sourced SynthID. Simultaneously, Meta introduced VideoSeal,151 a watermarking system for AI-generated videos. Leading AI companies are now increasingly integrating watermarking into their platforms. Google, for instance, is embedding SynthID into AI-generated images, text and videos across its services. Meta is applying invisible watermarks and metadata tags152 to AI-generated content on Facebook, Instagram and Threads. AI companies are partnering with organizations like Partnership on AI153 to ensure “synthetic media transparency”. Despite progress, though, widespread use of AI watermarking faces challenges.154 Simple modifications to AI-generated outputs can still disrupt detection. Users can attempt to remove or forge watermarks, either by cropping images and video where watermarks are embedded in a specific location, or by adjusting text (and even using AI- based watermark removers). Uneven adoption also presents risks where, without universal industry standards, inconsistent implementation may weaken effectiveness. There are also substantial ethical concerns around misuse, such as falsely labelling real content as AI-generated or false positives, where erroneous accusations of covertly using AI can have unintended consequences, especially in cases related to academic integrity. To be successful, these technologies will need to be accompanied by equally sophisticated governance and use guidelines. China has acted to regulate generated content to require watermarking,155 and other regions, such as the EU,156 are also developing responses to manage the security and authenticity of digital content. The Coalition for Content Provenance and Authenticity (C2PA), a coalition of leading media generators in the AI space, is also leading the development of technical standards for certifying the source of media content; an approach that regulators would struggle to meaningfully produce. Watermarking has proven a fertile area for start-ups globally with different technological approaches. Emerging generative AI watermarking technologies are becoming a cornerstone of responsible AI deployment as they help balance innovation with accountability. While no single method is foolproof, industry-wide adoption and regulatory alignment will help determine the technology’s long-term utility and success of AI-generated content.Katherine Daniell Director and Professor, School of Cybernetics, Australian National UniversityAndrew Maynard Professor, School for the Future of Innovation in Society, Arizona State University Boxed Icons Develop tamper-resistant watermarking standards – Invest in advanced watermarking technologies that can withstand removal attempts and establish industry-wide standards for their implementation. Ecosystem readiness map KEY ACTIONS TO ACHIEVE SCALE TechnologicalSocial EconomicEnvironmentalPolicy Image: Generative watermarking embeds invisible markers in content to verify authenticity, trace origins and promote accountability. Credit: Midjourney and Studio Miko. Prompt (abbreviated): “Digital fingerprint constructed from glowing binary code.” Read more: For more expert analysis, visit the generative watermarking transformation map. Authored by: Sri Krishnan.Boxed Icons Create cross-platform verification systems – Build independent verification systems that can detect and authenticate watermarks across different platforms and content types. Top 10 Emerging Technologies of 2025 37