Amazon joins C2PA steering committee to identify AI-generated content
As deepfakes and synthetic media become more common thanks to generative artificial intelligence (gen AI), several organizations are trying to perfect digital watermarking to determine a piece of content’s origins. None have really succeeded yet, but one such group, the Coalition for Content Provenance and Authenticity (C2PA) is beefing up its ranks.
On Friday, C2PA announced that Amazon has become one of the organization’s steering committee members. Part of the Linux Foundation family, C2PA describes itself as an “open, technical standards body addressing the prevalence of misleading information online through the development of technical standards for certifying the source and history (or provenance) of digital content.”
Also: IBM will train you in AI fundamentals for free, and give you a skill credential – in 10 hours
Adobe co-created C2PA and helped establish Content Credentials, which C2PA now maintains as one approach to making the origins of digital content more transparent. Content Credentials reveal how and when a piece of content was generated or modified. The coalition hopes to make Content Credentials an industry-wide standard.
By joining the coalition, Amazon agrees to link Content Credentials to content generated by Titan Image Generator v1 and v2, its enterprise AI image creator.
“With this Content Credentials implementation, consumers will be able to verify the origin and authenticity of the AI-generated content.”
Also: McAfee unleashes AI deepfake audio detector – but how reliable can it be?
The company will also add the provenance tool to AWS Elemental MediaConvert, “a file-based video processing service that transcodes content for broadcast and multi-screen delivery at scale,” the release explains. This means news organizations, sportscasters, aggregators, and other clients will be able to vet where the videos they’re sharing came from before further distributing them.
“All Amazon Titan-generated images contain an invisible watermark by default, providing a discreet mechanism to help reduce the spread of misinformation,” said Vasi Philomin, vice president of gen AI at AWS. “Joining the C2PA and adopting Content Credentials represents another key step in our journey towards responsible AI innovation.”
“The implementation of Content Credentials into tools and services that reach both news organizations and consumers at large further continues the C2PA’s mission of promoting transparency in the digital media ecosystem,” said Andrew Jenks, Executive Chair for the C2PA. Ideally, standardizing Content Credentials would reduce the amount of misinformation floating around and confusing consumers.
Also: Most people worry about deepfakes – and overestimate their ability to spot them
Philomin added that Amazon plans to “develop meaningful tools” for AI safety in collaboration with other C2PA members, which include Google, Meta, OpenAI, Microsoft, BBC, and Intel, among others.
That star-studded lineup, however, is an inherent issue for some.
In July, digital content certifier Numbers Protocol, alongside several other blockchain and AI organizations, founded the Creative Origin Alliance, a “democratic, transparent, and decentralized alternative” to C2PA in which “all members will have an equal voice, and creators will have a seat at the table.”
Also: Google upgrades Search to combat deepfakes and demote sites posting them
The Alliance was prompted by an Adobe Terms of Service update that suggested users’ work may be used to train Firefly, Adobe’s AI image tool, which made the digital creative community feel “preyed upon by big tech.” Concerned about conflicts of interest between C2PA and founders like Adobe, the Alliance aims to lift smaller voices in the industry, advocating for privacy and open-source tech, positioning itself as David to C2PA’s Goliath.
