Google signs EU AI Act transparency code of practice, arguing for practical AI provenance tools

Google is signing a new European voluntary code of practice focused on transparency for AI-generated content, an effort the company says is meant to make it easier to understand where online material came from and how it was created or edited.

In a post on Google’s public policy blog dated July 24, the company said it will back the EU AI Act Code of Practice on Transparency of AI-Generated Content. Google framed the move as part of a broader transparency push in Europe, building on its 2025 signing of the GPAI Code of Practice.

Google tied its commitment to technical steps it is pursuing to label and verify content provenance. The company said it is working to adopt and accelerate the C2PA industry standard, which is designed to help embed creation and editing metadata into digital content. It also pointed to SynthID, described by Google as its digital watermarking technology that helps people understand how content was created and edited.

To support what Google called “trust across platforms,” it said it is partnering with third-party AI labs including Apple, Eleven Labs, Kakao, NVIDIA, and OpenAI to drive adoption of interoperable watermarking tools using SynthID. Google presented that interoperability effort as a way to reduce fragmentation, so that transparency mechanisms work more consistently across different services and publishers.

While endorsing transparency, Google also warned that additional regulatory complexity could undercut the EU’s goals for competitiveness and simplification, and could confuse the people the rules are intended to help. The company said overlapping AI labels and legal disclosures could make it harder for users to get clear context about content they encounter online.

Google’s post added that its view is that privacy and security guardrails should not be weakened by implementation choices, and it said it has repeatedly offered solutions intended to safeguard users while meeting compliance goals. As the code is implemented, Google said it will work with regulators and partners to build practical solutions that provide real transparency rather than raising noise for consumers.

The company’s transparency messaging is part of a larger pattern in the AI industry, where policymakers are seeking ways to mitigate risks from synthetic media while the underlying tools are still evolving. By pointing to C2PA and watermarking interoperability, Google is effectively aligning itself with an approach that treats provenance as an ecosystem problem, not a single company or platform feature.

Google also mentioned an unrelated Europe-focused effort under, its philanthropic arm. The post said is providing a 5 million dollar grant to scale Obrii, Ukraine’s national AI job platform, describing it as contributing to Ukraine’s AI adoption.

Still, the post did not spell out how the code will be translated into enforceable requirements, what exact disclosure formats regulators will expect, or how different watermarking and metadata approaches will be reconciled at scale. It also did not provide timelines, benchmarks, or specific compliance details beyond its stated commitment to work with regulators and partners. The outcome may hinge on implementation guidance, industry coordination, and whether interoperability can keep pace with user-facing labeling requirements.