With the rise of advanced tools that enable the rapid creation, alteration, and distribution of digital content, there are many ways to manipulate what people see and believe. This guide observes the rapid uptake and evolution of AI and machine learning tools, including generative models and deepfake technologies, is outpacing traditional verification methods. As the abuse of AI-generated media can pose a significant cybersecurity threat, there is an urgent need to bolster information integrity.

This joint publication outlines leading approaches to confirming the provenance of digital content, with a focus on Durable Content Credentials. 

Key points

• The amount and quality of synthetic data is growing rapidly and will soon be indistinguishable from real content.
• Content provenance solutions aim to establish the lineage of media, including its source and editing history over time.
• These technologies can be used across different types of media, not just images, including video, audio and text.
• Durable Content Credentials are a provenance solution that uses cryptographically signed metadata describing the provenance of media. This metadata can be attached to the media content during export from software or at the point of creation.

Key recommendations

• Organisations should adopt Durable Content Credentials to provide transparency about their media content.
• Organisations should consider the stage at which to incorporate Content Credentials, including at the point of capture, during editing, or before publishing.
• Organisations should securely store media for verification purposes, preferably in read-only format.
• Policymakers should consider implementing laws that support transparency in the use of generative AI content.