Google Signs EU's AI Content-Labeling Code, but Pushes Back on Overlap
Google is signing the EU's AI transparency code, while warning that overlapping AI labels could confuse the people they aim to help.
What Google is signing
Google announced it is signing the EU AI Act Code of Practice on Transparency of AI-Generated Content, a voluntary commitment aimed at making it clearer to people in Europe when content they encounter was created or altered by AI. In its announcement, Google frames the move as reinforcing its commitment to transparency and responsible AI development in Europe, and ties it to work it says is already underway: adopting the C2PA industry provenance standard and building out AI transparency tools, including SynthID, Google's own digital watermarking technology. The code sits inside the broader EU AI Act framework, and signing it is a voluntary, public commitment rather than a legal requirement imposed directly on Google.
How this connects to Google's earlier commitments
This is not Google's first transparency pledge under the EU's AI rules. The company points back to its 2025 signing of the separate GPAI Code of Practice, positioning the new transparency code as a continuation of that earlier commitment rather than a standalone step. Google also says it has been partnering with other AI labs, including Apple, Eleven Labs, Kakao, NVIDIA, and OpenAI, to push for shared, interoperable watermarking tools built around SynthID, rather than each company shipping its own incompatible labeling scheme. The goal, as Google frames it, is for a watermark applied by one company's tool to still be readable by another company's detector, so a label does not simply stop working the moment content moves across platforms.
The catch: Google's own warning about the code
Alongside the signing, Google raises a pointed concern about the code it is agreeing to follow. The company argues that adding more regulatory complexity, while the underlying technical solutions for watermarking and detection are still evolving, risks working against Europe's own goals for competitiveness and simplification. Google also warns the added complexity could confuse the people the rules are meant to help, saying that if online content ends up covered in overlapping AI labels and legal disclosures, it becomes harder for people to get the clear context they need. In effect, Google is agreeing to the rulebook while publicly asking regulators not to let separate, overlapping labeling regimes pile up on top of each other. Google says it intends to keep working with regulators and partners as the code is implemented, aiming for practical solutions rather than a stack of disconnected disclosures.
What Google says happens next
Google frames this as an ongoing process rather than a finished commitment. The company says it will work together with regulators and partners as the code is put into practice, with the stated goal of building solutions that give people real, usable transparency about AI-made content rather than a checklist of labels nobody reads. Because the underlying detection technology is still maturing, how well this actually works in practice will depend on decisions still to be made by regulators and by the companies signing on alongside Google.
Key facts
- Google is signing the EU AI Act Code of Practice on Transparency of AI-Generated Content.
- The move builds on Google's earlier 2025 signing of the GPAI Code of Practice.
- Google links its commitment to the C2PA provenance standard and its own SynthID watermarking technology.
- Google has been partnering with Apple, Eleven Labs, Kakao, NVIDIA, and OpenAI on interoperable watermarking tools.
- Google separately warns that overlapping AI labels and disclosures could confuse users if the rules are not implemented carefully.
Check your understanding
- Which existing Google technology does the company say its EU transparency commitment builds on?
- What concern does Google raise about how the AI Act's transparency code could be implemented?