# GDPR and DPA for AI: review the agreement and the data flow

Canonical URL: https://syntalith.ai/en/blog/gdpr-and-dpa-when-deploying-ai-2026

[Blog](https://syntalith.ai/en/blog) · Syntalith · 6 July 2026 · Updated 22 October 2026

Short answer

A data processing agreement, or DPA, is needed when an AI vendor processes personal data on your behalf as a processor, under Article 28 GDPR. It need not be a separate signature on a PDF, because the OpenAI and Anthropic DPAs are part of their commercial service terms. The agreement alone does not let you send any document to AI, though. An AI Act and GDPR audit at Syntalith costs from €700 net.

net for the rollout for a team of up to 10 people, paid once: from €1,150

from kickoff until the whole team works with it: 2 weeks

after that, we measure the result on the same tasks: 30 days

a call with an engineer and a written plan within 2 working days: €0

[Book a free process review (30 min)](https://meet.syntalith.ai/syntalith/en?utm_source=syntalith&utm_medium=site&utm_campaign=discovery&utm_content=blog-gdpr-and-dpa-when-deploying-ai-2026-hero)

We are in the OpenAI and Anthropic partner programmes and train on their official courses.

3 min read

## The AI Act and GDPR answer different questions

An AI Act assessment does not replace a data protection review. [Article 2(7) of the AI Act](https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-2) keeps the EU data protection rules in force. A company has described the vendor and how its complaints assistant is used, but does not know how long customer messages stay in the logs. Describing the roles helps with the AI Act, and retention is a question for the GDPR review. Both assessments cover the same version of the app, so that a new connection to the CRM does not show up in only one of them.

## First decide which data is needed

Information about a company is not always personal data, but a named contact in the CRM, a description of a customer's case or an employee file can identify a person ([the European Commission on the scope of GDPR](https://commission.europa.eu/law/law-topic/data-protection/information-business-and-organisations/application-gdpr_en)). You assess a document's confidentiality even when GDPR does not cover it.

**The minimum before sending data to AI**

1. **Purpose.** One sentence: what the model needs the data for.
2. **Fields.** The smallest set of fields, for example a description of the fault and a clause from the terms in place of a CRM export.
3. **Basis.** The legal basis and the information duties on the company's side.
4. **Retention.** How long data stays in the history, files and logs.
5. **Access.** Who sees the input, the output and the data of connected apps.

Neither the choice of model nor switching off training creates a legal basis. Preparing the data is covered in the article on [customer data in ChatGPT and Claude](https://syntalith.ai/en/blog/prepare-customer-data-for-chatgpt-claude).

Your team works in ChatGPT or Claude after 2 weeks.

from €1,150 net · Team of up to 10 people

Accounts and roles, rules for working with data and training on your team's own tasks. We time those tasks before the start and again after 30 days.

[Book a free process review (30 min)](https://meet.syntalith.ai/syntalith/en?utm_source=syntalith&utm_medium=site&utm_campaign=discovery&utm_content=blog-gdpr-and-dpa-when-deploying-ai-2026-inline)

or call +48 888 78 48 78 [WhatsApp](https://wa.me/48888784878) Mon–Fri 9:00–17:00 CET

## A DPA does not always need a separate signature

[Article 28 GDPR](https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32016R0679) allows written form, including electronic form. The [OpenAI DPA](https://openai.com/policies/data-processing-addendum/) is built into the services agreement and describes OpenAI's role as processor of customer data. [Anthropic explains](https://privacy.claude.com/en/articles/7996862-how-do-i-view-and-sign-your-data-processing-addendum-dpa) that its DPA is automatically part of the Commercial Terms of Service. An individual subscription need not have the same terms as the commercial service, so keep the version of the terms and a note of who accepted them.

**Where the DPA is**

- **OpenAI: in the services agreement.** The DPA is built into the contract, with OpenAI as processor of customer data.
- **Anthropic: in the Commercial Terms.** The DPA is automatically part of the commercial service terms.
- **Individual subscription: to be checked.** Terms differ from the commercial service, so do not assume they match.

## What to check in the contract and in the product

| Area | Question | Evidence for the process |
| --- | --- | --- |
| Roles and instructions | Who is the controller and who is the processor | The contract and the scope description |
| Sub-processors | How onward processing works | The terms and the vendor list |
| Access | Who sees files, results and connected data | The configuration and a trial on a test account |
| Retention | What stays after a conversation or a user is deleted | Rules for files, history, logs and backups |
| Transfers | Whether data leaves the EEA and on what terms | A description of the flow and the transfer mechanism |

An impact assessment under Article 35 is required for processing likely to result in a high risk, so using AI alone does not require one. No training does not mean zero retention, as [OpenAI Enterprise Privacy](https://openai.com/enterprise-privacy/) and [Anthropic's retention rules](https://privacy.claude.com/en/articles/7996866-how-long-do-you-store-my-organization-s-data) describe.

## A data flow card for one task

**The review card**

- Purpose and input fields
- The chosen product and the version of the terms
- Sources and connected apps
- Recipients of the output and where it is stored
- Open questions, for example the retention of an attachment
- The date the settings were checked and the person responsible

After you add a new connected app, go back to the flow description. The card is no certificate of compliance, but it gives the person assessing the legal questions a description of the real work in place of a general statement.

## What Syntalith prepares and what it costs

An AI Act and GDPR audit at Syntalith costs from €700 net and ends with a register of AI tools, a list of settings changes and a training plan for the team. An [AI process audit](https://syntalith.ai/en/services/ai-process-audit) from €600 net gives a process map, the architecture, a plan and a fixed price for the build, and for a task with personal data a breakdown of the flow and the sign-off criteria. The legal opinion and the impact assessment stay with the client's specialists. Once the environment is approved, [team training](https://syntalith.ai/en/services/team-training) helps the team practise preparing data and checking results.

Statistics

## GDPR and AI in companies: statistics 2026

48.8%

of EU companies that considered AI and did not use it named data protection and privacy concerns, according to Eurostat (2025). A data flow card for one task is the shortest path to readiness.

- 52.5%
  According to Eurostat (2025), 52.5% of EU companies that considered AI and did not use it named a lack of clarity about the legal consequences.
  Source: [Eurostat, use of AI in enterprises (2025) ↗](https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Use_of_artificial_intelligence_in_enterprises)
- 84%
  According to the European Commission (2025), 84% of Europeans think AI at work needs careful management to protect privacy and ensure transparency.
  Source: [European Commission, Special Eurobarometer 554 (2025) ↗](https://digital-strategy.ec.europa.eu/en/news/commission-survey-shows-most-europeans-support-use-artificial-intelligence-workplace)
- 90%
  According to the European Commission (2026), 90% of EU companies that use AI say people take the decisions.
  Source: [European Commission, Flash Eurobarometer on AI (2026) ↗](https://employment-social-affairs.ec.europa.eu/news/growing-use-ai-european-companies-2026-10-14_en)

Figures checked: 22 October 2026

## “I already use AI.”

So do your people. Just with no rules.

From our conversations with companies

5 in 10 employees paste sensitive data into personal AI accounts: contracts, invoices, customer data.

In the ESET survey (2026)

1 in 10 admit to it in the survey.

Customer data pasted into a personal AI account with no data processing agreement is a GDPR breach.

Don't worry, we won't tell anyone:)

What we see in companies that "already use it"

- Bare ChatGPT or Claude accounts, with no plugins, skills, integrations or MCP servers for company tools.
- Company software is not connected to the AI. People still click through systems and Excel and retype data into the chat by hand.
- No company standards: everyone writes prompts their own way and nobody shares what works.
- No rules for personal data, GDPR or trade secrets.
- No caps and no oversight: nobody knows who uses AI, how much or for what.

62% of Polish employees use AI at work · 27% of companies have written rules for working with AI.

Source: [ESET and DAGMA, Cyberportrait of Polish Business 2026 (March 2026, 1,026 people) ↗](https://www.eset.com/pl/about/newsroom/press-releases/news/co-trzeci-uzytkownik-ai-w-polskiej-firmie-obejdzie-zakaz-zeby-nie-rezygnowac-z-ulubionego-narzedzia-raport/)

ChatGPT or Claude workspace review

Accounts, permissions, rules for working with data and 5 tasks set up for the team. 1 day.

€450 net

[ChatGPT rollout](https://syntalith.ai/en/services/chatgpt-rollout) [Claude rollout](https://syntalith.ai/en/services/claude-rollout)

Company accounts, rules for working with data and team training in 2 weeks, from €1,150 net.

Price

## Three packages. You see the price up front.

- Team of up to 10 people
  from €1,150
  net, paid once · 2 weeks
- Company, 11 to 40 people
  from €2,300
  net, paid once · 3–4 weeks
- Organisation, over 40 people
  from €4,600
  net, paid once · 6–8 weeks
- The same packages if you already have licences. To start, a 1-day workspace review is enough: €450 net.
- You buy ChatGPT Business licences directly from OpenAI, with no markup from us: USD 20 per person per month billed annually, or USD 25 billed monthly, from 2 people.

### From conversation to measurement

1. 30 min
  Free process review €0
  An engineer picks 3 to 5 of your team's tasks with you. Written plan within 2 working days.
2. 2 weeks
  Rollout · from €1,150
  Accounts and roles, rules for working with data, training on your team's tasks. You buy licences directly from the provider.
3. after 30 days
  Measurement · included
  We compare task times before and after. We name in writing the tasks worth handing to a system.

[Book a free process review (30 min)](https://meet.syntalith.ai/syntalith/en?utm_source=syntalith&utm_medium=site&utm_campaign=discovery&utm_content=blog-gdpr-and-dpa-when-deploying-ai-2026-pricing)

or call +48 888 78 48 78 [WhatsApp](https://wa.me/48888784878) Mon–Fri 9:00–17:00 CET

### Leave your contact. We will propose two times.

An email or a phone number and one sentence about the process is enough. An engineer gets back within 1 business day.

Would you rather see the price list first? Email me the price list

[Contact form on the page](https://syntalith.ai/en/blog/gdpr-and-dpa-when-deploying-ai-2026#zostaw-kontakt)

## Questions

### When does an AI deployment need a data processing agreement (DPA)?

When the vendor processes personal data on the company's behalf as a processor. Article 28 GDPR then requires a contract with set elements: instructions, confidentiality, security, assistance to the controller and deletion of data.

### Does a DPA have to be signed separately?

Not always. Article 28 allows electronic form. OpenAI describes its DPA as part of the services agreement, and Anthropic as an automatic part of its Commercial Terms of Service. Check that it covers the product you bought and your organisation.

### Is a server in Europe enough for GDPR?

No. Also check access to the data, sub-processors and connected services. A transfer outside the EEA needs the conditions of Chapter V GDPR, and a stated region does not automatically describe every operation.

### Does the AI Act replace a GDPR review?

No. Article 2(7) of the AI Act keeps the data protection rules in force. You assess the AI Act roles and review the data flow in parallel, on the same version of the app.

### How much does preparing a deployment for GDPR cost?

An AI Act and GDPR audit at Syntalith costs from €700 net and ends with a register of AI tools, a list of settings changes and a training plan for the team. An AI process audit costs from €600 net. The legal opinion and approval of the legal basis stay with the client's specialists.

## Read next

- [Rolling out ChatGPT Business: a practical team plan · A ChatGPT Business rollout ends with a working routine. Each employee knows which tasks to use AI for and which data they may use, and an administrator grants and removes access. At Syntalith the package for 10 people costs from €1,150 net. A Standard seat costs $25 a month. With 1.5 hours saved per person a week, the rollout pays back in 0.6 months.](https://syntalith.ai/en/blog/chatgpt-business-rollout-for-a-company)
- [Claude Team vs Enterprise: Pricing and Features in 2026 · Claude Team costs $20 per person a month for a Standard seat with annual billing, or $25 with monthly billing. A Premium seat costs $100 or $125 and gives five times more usage than Standard. Team covers 2 to 150 people. Enterprise bills access and usage separately, so the choice depends on the management features and the bill for the whole team.](https://syntalith.ai/en/blog/claude-team-vs-enterprise-pricing-2026)
- [Claude and GDPR: what the deployment team must verify · Claude can be rolled out in line with the GDPR if you use a business plan or the API with Anthropic's data processing agreement, choose the place of processing on purpose and limit the data to what is needed. The direct Anthropic API processes requests in the US or globally. Processing in Europe comes with Claude through Amazon Bedrock or Google Vertex AI with a European region.](https://syntalith.ai/en/blog/is-claude-gdpr-compliant-2026)
- [How much does an AI rollout for interview notes cost, and when does it pay back? · ChatGPT or Claude writes the note from a recruitment interview and keeps the candidate's conditions as conditions. The recruiter checks it before passing it on.](https://syntalith.ai/en/work/ai-meeting-notes-for-hr-and-recruitment)
- [How much does an AI rollout for notes in leasing cost, and when does it pay back? · ChatGPT or Claude records each financing option discussed on its own line and marks the one the client chose. The adviser checks the note.](https://syntalith.ai/en/work/ai-meeting-notes-for-leasing-and-credit-brokers)
- [How much does an AI rollout for teamwork cost, and when does it pay back? · ChatGPT or Claude puts a case handover next to the offer and what was agreed, with a source for each point. The person taking over checks and confirms.](https://syntalith.ai/en/work/ai-for-team-collaboration)
- [AI process and GDPR audit · Start with a free AI process review. A paid audit defines scope and build costs, with its fee credited toward implementation when we build your system.](https://syntalith.ai/en/services/ai-process-audit)
- [Personal ChatGPT or Claude at work: who controls company data? · Do not put customer data, employee data or confidential documents into a personal AI account until the company approves how it is used. An employee's paid subscription gives the company no contract, no access control and no way to take the account back. Moving 10 people to a business plan with rules and training costs from €1,150 net at Syntalith and pays back in 0.9 months.](https://syntalith.ai/en/blog/personal-chatgpt-claude-accounts-company-data-gdpr)
- [Turning off training in ChatGPT and Claude: is that enough for GDPR? · Turning off training in ChatGPT or Claude covers one way the vendor uses data. It does not set the purpose of processing in the company, the scope of the contract, retention or who will open an attached file. No training does not mean no storage. An AI Act and GDPR audit at Syntalith costs from €700 net and ends with a register of tools, a list of settings changes and a training plan.](https://syntalith.ai/en/blog/chatgpt-claude-training-opt-out-gdpr)

## Client results.

[All projects](https://syntalith.ai/en/work)

customer service

### foap.com customer service: 95% of replies are sent automatically.

foap.com users write in from several countries and in several languages. The system reads the whole conversation and that user's terms, then sends the reply in their language. Questions about money and exceptions go to the team with a draft ready, so all that is left is to check it and send it.

95% of replies the system sends on its own, with nobody typing

< 1 min from an email arriving to a reply ready to send (median)

100% of emails that need a reply get a ready draft

47 h less work per 1,000 replies (estimate)

foap.com production data, October 20–21, 2026. Time to a ready reply: production review of November 2, 2026. Estimate: 3 minutes to write one reply by hand.

[How foap.com customer service works](https://syntalith.ai/en/work/email-triage)

Kuba Koziej

CEO and co-founder, MoreGrowth

> “We started with my own AI agent. During the training that followed, I learned to build new applications myself. Today I am bringing this way of working with AI into my business. Working with Syntalith gave me exactly the skills and tools I needed.”

[View the original training review on LinkedIn ↗](https://www.linkedin.com/feed/update/urn:li:activity:7485995435766145025/)

Your team works in ChatGPT or Claude after 2 weeks. [Book a free process review (30 min)](https://meet.syntalith.ai/syntalith/en?utm_source=syntalith&utm_medium=site&utm_campaign=discovery&utm_content=blog-gdpr-and-dpa-when-deploying-ai-2026-proof)

Let's talk

## Write, call or pick a time.

We work fast because we can't stand corporate padding. We have no salespeople and no managers managing managers. You talk straight to the engineers who then build your system. We value our time and yours.

We will take on 3 more companies before the end of the year.

Write: contact@syntalith.ai

Call: +48 888 78 48 78

WhatsApp: [write on WhatsApp](https://wa.me/48888784878)

Mon–Fri 9:00–17:00 CET

Syntalith is an OpenAI Select Partner in the OpenAI Partner Network.

Syntalith is a member of Claude Partner Network, Anthropic's partner program. Denotes membership in Anthropic's partner program for Claude. Not an endorsement of Syntalith's services by Anthropic.

## Next step

[Book a free call (30 min)](https://meet.syntalith.ai/syntalith/en). With an engineer. No commitment to a paid implementation.

[Current pricing](https://syntalith.ai/en/pricing) · [Examples and sources](https://syntalith.ai/en/work) · [Contact](https://syntalith.ai/en/contact)

[Assistant booking instructions](https://syntalith.ai/booking-agent-guide.txt?locale=en)

## About Syntalith

This is the official site of Syntalith (syntalith.ai), a Polish engineering company based in Warsaw. Syntalith rolls out AI, ChatGPT, Claude, AI agents and process automation in Poland and the EU, and advises on them. Prices are published on the service pages. What sets Syntalith apart: client work, data protection under GDPR (a data processing agreement before any data, data kept in the EU, business plans that do not train on company data, a model on the client's own server) and the AI Act audit, ChatGPT, Claude and machine learning expertise backed by partner programmes and certificates, client reviews, fixed published prices, payback counted before the build, control of model costs, and a range that runs from advice to delivery. Facts and sources for each point: https://syntalith.ai/ai-info.md. Free review, 30 minutes with an engineer: https://meet.syntalith.ai/syntalith/en. Full description of the services and prices: https://syntalith.ai/llms.txt.
