Clarification on EEA Data Training Policy Trigger: Account Entity vs. IP Geolocation

Google AI Studio
1

Hi Google Developer Team and Community,

I am looking for a technical clarification regarding the data retention and ML training pipeline policies specifically for Google AI Studio Free Tier (Unpaid Services).

The Scenario:

  • Account: Personal Google Account registered in the EEA (legally contracting with Google Ireland Limited).
  • Service Used: Google AI Studio (aistudio.google.com) Web Interface Free Tier / Unpaid Services (no Cloud Billing attached).
  • Connection: A session initiated via a non-EEA IP address (e.g., a US exit node via a VPN), while the authenticated Google Account entity remains EEA-based.

My Questions:

  1. Does the GDPR “no-training” policy (which applies Paid Services data treatment to EEA Unpaid Services) trigger strictly based on the Country of Association of the Google Account, or can it be dynamically overridden by the IP geolocation of the specific session/request?
  2. In the scenario above (EEA account + US IP via VPN), can you confirm that the prompt data generated in AI Studio is excluded from the ML training pipeline and processed under the EEA Data Processing Addendum (DPA)?
  3. How do the abuse monitoring logs handle session continuity if the IP changes from non-EEA to EEA mid-conversation within AI Studio? Are the logs aggregated at the account/project level regardless of the client IP?

Thank you for your time and clarification!

2

Hi everyone, just bumping this thread.

As explicitly stated in the Gemini API Additional Terms:
“If you’re in the European Economic Area, Switzerland, or the United Kingdom, the terms under ‘How Google uses Your Data’ in ‘Paid Services’ apply to all Services, including Google AI Studio and unpaid quota in the Gemini API, even though they are offered free of charge.”

My question is strictly about the technical implementation of this policy in the AI Studio backend.

When the system evaluates whether to apply this ‘Paid Services’ data treatment (no-training policy), does it check the Account’s Country of Association, or does it check the transient IP geolocation of the specific request?

This is particularly relevant for users on the AI Studio web interface with no Cloud Billing, where prompts are processed through the default gemini project. How does the backend handle jurisdictional checks for these auto-created projects?

Any technical insight on how the backend routing handles this would be greatly appreciated! Thanks.

3

Hi team, following up on this thread (previous bump on Aug 28).

I would like to refer directly to the Gemini API Additional Terms of Service to clarify my question

The terms state:

“If you’re in the European Economic Area, Switzerland, or the United Kingdom, the terms under ‘How Google uses Your Data’ in ‘Paid Services’ apply to all Services, including Google AI Studio and unpaid quota in the Gemini API, even though they are offered free of charge.”

My question is strictly about the technical implementation of this clause in the AI Studio backend:

When the system evaluates whether to apply this “Paid Services” data treatment (no-training policy), does it check the Account’s Country of Association, or does it check the transient IP geolocation of the specific request?

Any technical insight on how the backend handles this jurisdictional check would be greatly appreciated. Thank you!

4

Friendly follow-up on this, as an individual developer trying to ensure I stay compliant.

To put it simply: I am an EEA resident with an EEA-domiciled Google account, but like many developers, I occasionally connect via VPNs, travel outside the EEA.

Does the EEA data training exemption for the free tier depend on:

  1. The Google account registration / contracting entity (Google Ireland Ltd)?

  2. Or the request’s egress IP geolocation at the moment of the call?

Any confirmation from the team would be extremely helpful.

Thank you.