Current artificial intelligence detection tools face a major shortcoming when dealing with manipulated media. While they can successfully recognize specific traits left behind by known generative programs, they simply cannot verify the actual historical background of a digital file. We are excited to share a solution to this issue with OpenWater. Our proposal offers a totally open and decentralized system designed to tackle deepfakes and related problems by establishing clear digital provenance.
By utilizing resilient watermarking and universally compatible credentials, OpenWater ensures that a piece of media stays firmly connected to its historical record. This vital connection persists even in cases where standard metadata has been deleted. To achieve this, the platform actively logs signed statements that detail exactly how a file was originally created, altered, and subsequently shared with the public.
Rather than relying on a single central governing body, the OpenWater infrastructure operates across a wide network of independent services. This distributed design allows everyday users, organizations, and AI programs to assess the available evidence based on their own unique guidelines. Beyond just digital media, this robust framework is adaptable enough to cover the actions of autonomous agents, the outputs of various models, and the datasets used for training. As a result, artificial intelligence systems are granted direct access to the historical context of the processes and information they interact with.
Find out more about this initiative at https://bengoertzel.substack.com/p/toward-a-truly-decentralized-digital
By utilizing resilient watermarking and universally compatible credentials, OpenWater ensures that a piece of media stays firmly connected to its historical record. This vital connection persists even in cases where standard metadata has been deleted. To achieve this, the platform actively logs signed statements that detail exactly how a file was originally created, altered, and subsequently shared with the public.
Rather than relying on a single central governing body, the OpenWater infrastructure operates across a wide network of independent services. This distributed design allows everyday users, organizations, and AI programs to assess the available evidence based on their own unique guidelines. Beyond just digital media, this robust framework is adaptable enough to cover the actions of autonomous agents, the outputs of various models, and the datasets used for training. As a result, artificial intelligence systems are granted direct access to the historical context of the processes and information they interact with.
Find out more about this initiative at https://bengoertzel.substack.com/p/toward-a-truly-decentralized-digital
