josecustom.ai josecustom.ai Book

GDPR and AI for US Businesses: When It Applies to You and What to Set Up

A plain-English guide for US small businesses with European customers: the two tests that decide whether the GDPR reaches you, what it means for the AI tools you use, a vendor checklist, and the setup steps that cover most of the obligation.

GDPR compliant AI means using AI tools in a way that meets the European Union’s General Data Protection Regulation whenever those tools process personal data about people in the EU. For a US small business, the first question is not which tool to buy. It is whether the GDPR applies to you at all, and for many US businesses the honest answer is “only partly, and only for some of your data.”

I work with businesses in both markets. In Germany, where I also run a practice, this is a daily conversation. In the US, it usually comes up the first time a European client sends over a data processing agreement and asks which AI tools touch their information. This post is the version I wish those US owners had read beforehand.

One note before we start: I am a security engineer, not a lawyer. What follows is how the regulation is generally understood and what I set up in practice. If you have significant EU exposure, have a privacy attorney confirm the details for your situation.

Does the GDPR apply to a US business?

It can, even if you have no office, employee, or server in Europe. Article 3(2) of the GDPR extends the regulation to businesses outside the EU in two situations:

  1. You offer goods or services to people in the EU. Paid or free. The test is whether you are aiming at them, not whether someone from Europe happened to find your website. Signs of aiming include pricing in euros, a site in a European language, shipping to EU countries, or marketing that names EU customers.
  2. You monitor the behavior of people in the EU. Tracking and profiling visitors from the EU, for example with analytics or advertising tools that build behavioral profiles, can bring you into scope.

If neither applies, the GDPR generally does not reach you. A plumbing company in Orlando whose only European contact is a tourist who once called for a repair is not the business this rule was written for.

If one does apply, the GDPR covers the personal data connected to that activity. It does not automatically pull in everything else you process. That is why the useful next step is a data inventory, not a panic.

The common US scenarios

  • You have EU clients or customers. A consulting firm with clients in Ireland, an online store that ships to France, a software company with German users. In scope for that data.
  • You are a vendor to an EU company. Your client is the one primarily responsible for its customers’ data, and it will pass obligations down to you in a contract. You are acting as what the GDPR calls a processor, and that contract is what you have to meet.
  • You have employees or contractors in the EU. Their personal data is covered.
  • You occasionally get a European inquiry. Usually not enough on its own, but worth a sentence in your privacy notice.

What changes when you add AI tools?

Nothing about the GDPR is specific to AI. The same rules apply to an AI tool that apply to your CRM or your email provider. What changes is how easily personal data ends up somewhere you did not plan for. An employee pastes a European client’s email thread into a chatbot to draft a reply, and personal data has just gone to a new vendor, possibly in a new country, possibly into a product whose terms allow training on it.

Five GDPR concepts do most of the work here.

Lawful basis (Article 6). You need a valid legal reason for each use of personal data. If you collected a client’s details to deliver a service, using an AI tool to help deliver that service is usually easier to justify than feeding the same data into an unrelated experiment.

Processor contracts (Article 28). When a vendor processes personal data on your behalf, the GDPR requires a written contract with specific terms, usually called a data processing agreement or DPA. Business tiers of the major AI products generally offer one. Free consumer accounts generally do not, and that alone usually rules them out for EU personal data.

International transfers (Chapter V). Personal data leaving the EU needs a recognized legal mechanism. For US vendors, the common ones are certification under the EU-US Data Privacy Framework, which the European Commission recognized in 2023, and Standard Contractual Clauses built into the vendor’s DPA. Check which one your vendor relies on, and check where its subprocessors are. I walked through how to find out where AI data physically goes in AI data residency.

Transparency (Articles 13 and 14). People have the right to know how their data is used and who receives it. If AI vendors process your clients’ personal data, your privacy notice should say so in plain words.

Automated decisions (Article 22). People have the right not to be subject to decisions based solely on automated processing that produce legal effects or similarly significant effects on them. If an AI tool decides on its own whether someone gets credit, a job interview, or a contract, that is a different and much stricter situation than an AI drafting an email a human reviews.

Which AI tools can you use with EU personal data?

There is no official list of “GDPR compliant AI tools.” Compliance depends on the contract, the configuration, and how you use the tool, not the brand name. The same product can be appropriate in a business tier with the right settings and inappropriate in a free personal account.

Before you allow any AI tool to touch EU personal data, get yes or no answers to these:

  • Is there a DPA, and have you accepted it? Many vendors publish one that applies automatically to business plans. Some require you to sign it. Save a copy.
  • Is your data excluded from model training? On business tiers this is often the default. Confirm it in writing, not in a marketing page.
  • Where is data processed and stored? Does the vendor offer EU data residency, and is it actually enabled on your account?
  • What transfer mechanism does the vendor rely on? Data Privacy Framework certification, Standard Contractual Clauses, or both.
  • Who are the subprocessors? Reputable vendors publish a list and notify you of changes.
  • How long is data retained, and can you delete it? You need to be able to honor deletion requests.
  • Can an admin see and control usage? You cannot meet obligations for a tool you cannot see.

I cover the broader version of this checklist, including security and lock-in questions, in AI vendor risk assessment.

What should a small US business actually set up?

Most of the obligation for a small business comes down to six steps.

1. Map where EU personal data lives

List the systems that hold data about EU clients, customers, or staff. Then list the AI tools your team uses. The overlap is your work list. If you do not know which AI tools your team uses, start there, because employees often adopt them without asking.

2. Move that data into approved business accounts

Pick the AI tools you will allow for client work, set them up on business plans with a DPA, turn off training on your data, and enable EU data residency where it is offered and you need it. Then say plainly that other AI tools are off limits for client data. Your AI acceptable use policy is where that rule goes in writing.

3. Keep the sensitive categories out

The GDPR gives extra protection to special categories of data in Article 9, including health information, racial or ethnic origin, religious beliefs, and biometric data used to identify someone. Unless you have a specific reason and a specific legal basis, keep these out of AI tools entirely.

4. Update your privacy notice

Say which categories of vendors process personal data, including AI services, and how people can contact you to exercise their rights. Plain language beats a wall of legal text.

5. Be ready for a rights request

EU individuals can ask for a copy of their data, a correction, or deletion. Know who handles the request and how you would find their data across your systems, including any AI tools that retain conversation history.

6. Check the less obvious obligations

Non-EU businesses in scope under Article 3(2) generally must appoint a representative in the EU under Article 27, with a narrow exception for occasional, lower-risk processing. If your AI use is likely to create a high risk for people, for example large-scale profiling, a data protection impact assessment under Article 35 may be required. These are the points where a privacy attorney earns their fee.

What happens if you get it wrong?

The GDPR allows fines of up to 20 million euros or 4% of worldwide annual turnover, whichever is higher, for the most serious violations. Those headline numbers come up in every article on the topic, and they are real, but they are a ceiling, not a typical outcome for a small business.

For most US small businesses, the practical consequence shows up earlier and closer to home: a European client sends a security questionnaire or a DPA, asks which AI tools touch their data, and expects a clear answer. A business that can answer in an afternoon keeps the client. A business that has to start investigating may not.

Frequently asked questions

Is ChatGPT GDPR compliant?

No AI product is compliant or noncompliant on its own. Compliance depends on the account type, the contract, the settings, and how you use it. Business and enterprise tiers of major AI tools generally offer data processing agreements and exclude your data from training by default, while free personal accounts generally do not and are a poor fit for EU personal data.

Does the GDPR apply to US companies?

It applies to US companies that offer goods or services to people in the EU, or monitor their behavior, under Article 3(2). It also reaches US companies indirectly when they process personal data as a vendor for an EU business, through the contract that business signs with them. A US company with no EU customers, users, or clients is generally outside its scope.

Do I need a DPA with my AI vendor?

Yes, if the vendor processes EU personal data on your behalf. Article 28 of the GDPR requires a written contract with specific terms. Most major AI vendors offer a DPA for business plans, sometimes accepted automatically through their terms and sometimes requiring a signature.

Can EU personal data be processed in the US?

Yes, if the transfer has a recognized legal mechanism. For US vendors, the most common are certification under the EU-US Data Privacy Framework and Standard Contractual Clauses included in the vendor’s DPA. Some vendors also offer to store and process data in the EU, which reduces the transfer question.

Is anonymized data covered by the GDPR?

Data that is truly anonymous, meaning nobody can reasonably identify the person from it, falls outside the GDPR. Data with names removed but still linkable to a person, such as an email thread with job titles and dates, usually still counts as personal data. When in doubt, treat it as personal data.

The one-sentence version

Check whether you offer services to or monitor people in the EU, inventory where their data lives, allow only AI tools with a signed DPA, training turned off, and a known transfer mechanism, and update your privacy notice to say so.

If you would rather have the AI side set up correctly from the start, with business accounts, training off, data location confirmed, and admin visibility in place, that is what I build in a secure AI work environment. It is designed so that when a European client asks which AI tools touch their data, you already have the answer.


Jose Lugo is a CISSP-certified security engineer with 12 years of U.S. Army intelligence experience. He builds secure AI work environments and fast, maintainable websites for businesses at josecustom.ai. See his portfolio of 13 live client systems at portfolio.josecustom.ai.