Two EU developments put different parts of the AI data lifecycle under scrutiny: access to data from connected devices and the privacy risks of realistic AI-generated imagery. For data teams, both call for clearer decisions about who can use data, under what conditions, and with what safeguards.
Data Act | Shaping Europe’s digital future
The European Commission says the Data Act has applied since September 2025. It gives users greater control over data generated by connected devices and is intended to support a fairer, more competitive European data economy. Connected devices placed on the EU market must be designed to allow data sharing, giving consumers more scope to choose services without relying on the device manufacturer.
For a device vendor, that makes data access a product-design question, not just a request for the legal team to handle. For an analytics provider, it raises practical questions about which device-generated data a user can obtain or direct elsewhere, how that data is delivered, and whether existing partner arrangements assume exclusive access. Teams should distinguish the Act’s stated objectives from the specific obligations that apply to their products and roles before changing a data pipeline.
- Connected-device vendors should map how users access and share device-generated data, then check whether the product experience supports those paths.
- Data teams should document which telemetry and product data they collect, where it flows, and which access controls govern downstream use.
- Service providers should revisit contracts and integrations that depend on the manufacturer being the sole route to device data.
- AI teams using connected-device data should verify that their sourcing and governance processes account for user-directed sharing.
EDPB backs privacy warning on AI-generated imagery
The European Data Protection Board supports a joint statement on AI-generated imagery and privacy coordinated by the Global Privacy Assembly’s International Enforcement Cooperation Working Group. The statement addresses AI systems capable of producing realistic images and videos of identifiable people without their consent. The concern is not whether an image is synthetic in a technical sense, but what happens when a generated depiction can be associated with a real person.
That distinction matters for teams that treat synthetic output as automatically outside a privacy review. A generated image may not be a photograph, yet its resemblance to an identifiable individual can still create privacy concerns. The EDPB’s support is a signal to examine output controls and misuse scenarios alongside training-data practices; it is not, by itself, a new rule or a finding against a particular product.
- Image and video product teams should test whether users can generate realistic depictions of identifiable people without consent and assess appropriate safeguards.
- Privacy reviews should cover generated outputs and their potential use, rather than assuming synthetic media cannot raise identity-related concerns.
- Compliance teams should track how data protection authorities address AI-generated imagery, while separating regulatory statements from binding requirements.
- Buyers of generative AI tools should ask vendors how they handle likeness-related misuse, reporting, and controls for realistic imagery.
