Ben Goertzel launches OpenWater decentralized media provenance protocol for deepfake verification

This protocol provides a technical alternative to centralized provenance systems like Google's SynthID by distributing credential storage across independent services. For the streaming industry, it offers a standardized grammar of evidence to combat the 'liar's dividend,' where authentic footage is dismissed as synthetic. As AI-generated content saturates distribution platforms, this decentralized approach allows publishers and viewers to apply custom trust policies rather than relying on a single corporate gatekeeper. The success of this framework now depends on broad adoption across hardware manufacturers and editing software suites to ensure the chain of custody remains intact from capture to consumption.

The Coalition for Content Provenance and Authenticity (C2PA) has become the de facto standard for media provenance, with major technology companies and hardware manufacturers building compliance into their products. In early 2025, Adobe announced that C2PA Content Credentials would be embedded directly into the metadata of images captured by Nikon and Leica cameras, extending the provenance chain from the moment of capture. The coalition's membership now includes over 400 organizations spanning camera manufacturers, social platforms, and news agencies, creating a broad ecosystem that any new provenance protocol must interoperate with to gain traction.

Google's SynthID represents the most commercially deployed alternative to C2PA-based approaches, and its recent expansion underscores the competitive pressure facing decentralized proposals like OpenWater. Google DeepMind expanded SynthID to cover AI-generated video and 3D models in late 2024, building on its earlier deployment for images and text. The watermarking system has been integrated into Google's Gemini models and Imagen, processing billions of pieces of content. Meanwhile, the U.S. Federal Communications Commission proposed rules in 2024 requiring disclosure of AI-generated content in political advertising, signaling that regulatory mandates may force provenance adoption regardless of which technical framework prevails. The EU AI Act, which entered into force in August 2024, similarly requires providers of generative AI systems to ensure outputs are detectable as artificially generated.

SingularityNET, the organization behind OpenWater, has been building decentralized AI infrastructure since 2017, but its track record in media provenance specifically is limited. SingularityNET partnered with the United Nations Development Programme in 2024 to explore decentralized AI governance frameworks, suggesting institutional interest in its governance model. The technical challenge for OpenWater lies in achieving the same watermark robustness that centralized systems like SynthID have demonstrated. A 2024 study from the University of Maryland found that invisible watermarks degrade significantly under common transformations such as compression, cropping, and resizing, a problem that any provenance system relying on embedded signals must solve before it can function as a reliable chain of custody across real-world distribution pipelines.