Skip to main content

Privacy and data processing

These are the answers procurement, legal and data protection officers ask for before an agreement: which roles we have under the General Data Protection Regulation (GDPR), how you get access to personal data, when you need a data processing agreement, where the data is stored, which sub-processors we use, and what happens when an AI tool fetches data from Placepoint.

Send any questions not answered here to support@placepoint.no.

Roles under the GDPR​

Is Placepoint a controller or a processor?​

Both, for different types of data.

  • For the property and company data we collect and pass on, including data from grunnboken (the land register), the matrikkel (the cadastre) and Brønnøysundregistrene (BRREG), we are an independent controller under GDPR Article 4(7). We decide the purpose and means of the collection, and you receive the data from us. This is not a processing assignment, and it does not give you the right to instruct us about the register data. Integration and security shows which fields in the delivery are personal data.
  • For data you enter in Placepoint yourself, such as your own datasets, projects, notes, tags and what you type in a search or lookup, you are the controller and we are the processor under Article 4(8). We process this data only to deliver the service to you.
  • For information about you as a user, such as your name, email address, phone number and login and usage data, we are the controller. This is described in the privacy policy.

For your part, you are the controller for how you go on to use the register data. This applies whether you fetch it in Fusion, through the Placepoint API or through an AI tool. If you pass it on to others, the rules in Disclosure of PII to third parties apply.

Why is the register data not a processing assignment?​

Because we do not process it on your behalf. We fetch grunnbok, matrikkel and company data from the sources, combine it and offer it to all our customers on the same terms. A data processing agreement for the register data would give each customer the right to instruct us about, and delete from, a shared dataset, and that is not possible. Instead, disclosure takes place under our agreement with Kartverket and within section 4 of the regulations on disclosure, reuse and other processing of information from grunnboken and the matrikkel. These regulations also require you, as the recipient, to have a legitimate need.

Access to personal data​

Personal data about private individuals is hidden in Placepoint until you have told us what you need it for, and we have approved it. This applies to all customers, including those of you who saw such data in Placepoint 1. Without access, private individuals are shown as Private person, while companies are always shown with their name and organisation number.

How do we get access?​

Send an email to support@placepoint.no with the following:

  1. What you will use the data for, as specifically as possible. For example, selling homes you have been instructed to sell, or credit assessment of loan applicants.
  2. The legal basis you rely on. For most customers, this is legitimate interest under GDPR Article 6(1)(f).
  3. Which data you need: owners, buyers and sellers in Norway, shareholders and role holders, or owners in Sweden.
  4. A confirmation that you will follow the data protection rules and use the data only for the purpose you have stated.

We assess the need and reply to your email. When the need is approved, we open access, and it applies to all users in your account.

Why do we have to justify the need?​

Because the rules for the data require a specific purpose, and we must be able to document it.

If you rely on legitimate interest, you must also have weighed your interest against the privacy of the people concerned. Datatilsynet (the Norwegian Data Protection Authority) explains the balancing test.

What do we get access to?​

Access has two levels:

  • Standard level: the names of private individuals in the Norwegian sources, that is grunnboken, the matrikkel, the shareholder register and Enhetsregisteret (the Central Coordinating Register for Legal Entities) from Brønnøysundregistrene (BRREG).
  • Extended access: owners in Sweden, data from Enova and national identity numbers. This requires a separate justification for those specific data. You get national identity numbers only when you need them to identify a person reliably, under section 12 of personopplysningsloven (the Personal Data Act).

Placepoint AI never shows owners who are private individuals or national identity numbers, whatever access you have. What names an AI tool can get otherwise is described in What personal data can an AI tool get from Placepoint?

The Agent in Fusion shows the same as the panels, according to the access you have. The names of property owners are still not sent to the language model, with one reservation, see Does the language model behind the Agent get names of private individuals?

What are we responsible for once access is open?​

You are the controller for how you use the data. This means:

  • Use the data only for the purpose you stated. A new purpose requires a new justification to us.
  • Do not use data from grunnboken or the matrikkel for advertising or marketing without consent from the person the data concerns. This is set out in the third paragraph of section 5 of the regulations.
  • Give the data subjects information about the processing under GDPR Article 14, with the exceptions Datatilsynet describes.
  • Do not store the data for longer than the purpose requires.
  • Let each person log in with their own account. See Account and login.

If a customer breaks the rules, we must close their access and notify the supplier of the data. For data from Kartverket, this follows from Kartverket's terms and section 6 of the regulations.

Which rules apply to owners in Sweden?​

Swedish owner data comes from Lantmäteriet. Section 2 of Lag (2000:224) om fastighetsregister (the Swedish Real Property Register Act) lists the purposes personal data from the real property register can be used for, including property transactions, credit and insurance assessments and property management. The purpose you state must fit one of them, and Lantmäteriet checks the purpose before it discloses personal data (Lantmäteriet: Anvisningar, del 2 fastighetsinformation, point 3).

IMY, the Swedish data protection authority, describes the same requirement: before personal data is passed on, whoever discloses it must find out why the recipient wants it, whether they really need it, and what it will be used for (IMY: Intresseavvägning).

Data processing agreement​

Do I need a data processing agreement with Placepoint?​

Yes, if you enter personal data in Placepoint. In practice you do this when you upload your own datasets with names or contact details, use Placepoint Dataset MCP with your own projects, order an extract where you send us personal data as a key, or type a national identity number in a lookup. We then process the data on your behalf, and GDPR Article 28 requires a written agreement before processing starts.

The personal data fields in the Placepoint API also require a data processing agreement. If you only use the register data and the map, you do not need a data processing agreement for that. Many customers sign one anyway because procurement asks for it, and that is fine. We then limit Appendix A to what we actually process: the user accounts, the login and the logs. The register data is kept outside the agreement.

What does the Placepoint data processing agreement contain?​

We have a template reviewed by lawyers, based on Article 28, with four appendices:

  • Appendix A: what is processed, which categories of data subjects it covers, and the purpose.
  • Appendix B: our sub-processors, with service and country.
  • Appendix C: the instructions for the processing.
  • Appendix D: change log.

The agreement limits the assignment to the data you enter yourself. The register data is outside its scope, for the reason described above. Ask your contact person or support@placepoint.no for the agreement, and we will fill in the appendices together.

Can I use my own data processing agreement?​

Yes. We are happy to fill in your template, and have done so for municipalities, banks and insurance companies. We ask for two things in any template: that register data from the matrikkel, grunnboken and the company register is delivered by us as an independent controller and is not covered by the agreement, and that the list of sub-processors is the one we publish in the docs, so it can be kept up to date without a new round of signatures.

If your template requires all processing to take place in the EU/EEA, or every transfer outside the EEA to rely on standard contractual clauses, that point must be adapted to the list. Sentry and HubSpot are in the USA under the EU-US Data Privacy Framework, and the AI features in Fusion use Anthropic in the USA.

Storage and operations​

Where is the data stored?​

In Microsoft Azure in the Norway East region, with a replicated copy in Norway West. This covers the register data, the map layers, the user database and any datasets you upload. The exception is Placepoint AI for Sweden, the server for Swedish property, which runs at Amazon Web Services (AWS) in Ireland. Some supporting services, such as error logging and the customer system, are run by suppliers with servers in the EU or the USA, and the AI features in Fusion (the Agent, map layer chat and filter chat) use language models from Anthropic in the USA. The list of sub-processors shows which suppliers these are, what they process and which transfer mechanism applies.

Does US law, such as the CLOUD Act, apply to the data?​

Yes, for the suppliers that are American companies. The CLOUD Act is a US law from 2018 that lets US authorities demand data from US cloud services, even when the data is stored in another country. Microsoft, which operates Azure where the platform and the data are hosted, is an American company. The same applies to most of the companies in the list of sub-processors, including Anthropic.

Against this are the commitments Microsoft has made in its data processing agreement, which Microsoft describes for public sector customers: Microsoft does not disclose data to authorities unless the law requires it, tries to redirect the authorities to us, notifies us with a copy of the demand unless the law forbids it, and can challenge the demand in court when it conflicts with the law of the country where the data is stored. The databases are encrypted at rest with Azure's standard encryption, with keys managed by Microsoft. We have no measures of our own against the CLOUD Act beyond this.

What personal data does Placepoint process about users?​

Name, email address and phone number linked to the account, which organisation the user belongs to, and logs of logins and usage. The usage log is what lets us document who has seen which personal data, which we are required to do for data from grunnboken. It is also why each person must log in with their own account.

How is the data secured?​

  • Each user has a personal account, and access to personal data is set per account after a specific assessment of the legal basis for processing.
  • Every login is logged with its IP address, and the terms of use require the service to be used from the EU/EEA. See Access outside the EU/EEA.
  • All traffic is encrypted with TLS, and the databases are encrypted at rest with Azure's standard encryption.
  • Production data is replicated to another Azure region, and access to it is limited to named operations staff.
  • Events in the platform are logged, so we can trace who has done what.

Technical security covers TLS versions, cipher suites, host names and backups. If procurement needs a completed security description or a supplier questionnaire, send it to support@placepoint.no.

What happens in a security breach?​

We notify you in writing within 48 hours of discovering a breach affecting data we process for you. We tell you what happened, which data is affected and what we are doing. This gives you a basis for your own assessment under GDPR Article 33. The contact point is given in the data processing agreement.

How long does Placepoint store the data?​

Your own datasets, projects and notes stay until you delete them yourself or the agreement ends. When the agreement ends, we delete or return the data as the data processing agreement specifies, and confirm this in writing. Information about users is stored for as long as the customer relationship lasts, and after that for as long as accounting and contract rules require.

Backups are not deleted when the agreement ends, but expire on a fixed schedule: point-in-time restore copies after seven days, weekly copies after six months, monthly copies after two years and annual copies after ten years. After that, the data is also gone from the backups. Technical security describes how the backups are stored.

Operations logs in Azure are deleted after 30 days, and application logs after 90 days. Lookups from AI tools are deleted after 90 days, see What does Placepoint log when an AI tool fetches data?

Placepoint AI and your own datasets​

Who is the controller when an AI tool fetches data from Placepoint?​

You are. Placepoint AI consists of MCP servers that answer lookups from the AI tool you have chosen, using your login and your access level. This applies to every server we offer, including future ones. The servers contain no language model, and we send no data to an AI provider. The AI tool fetches data from us on your behalf, and you decide which tool is used, which questions are asked and what the answers are used for. This makes you the controller for what happens in the AI tool, and your agreement with the AI provider governs storage, location and any training there.

This is not joint controllership under GDPR Article 26, because we decide neither the purpose nor the means of the processing in the AI tool. Our responsibility ends with what happens in our own systems.

Where does the data go when an AI tool uses Placepoint AI?​

The diagram shows the Norwegian servers. Placepoint AI for Sweden, the server for Swedish property, runs at Amazon Web Services (AWS) in Ireland, and lookups to it are processed there.

From us, the data goes back only to the tool that asked. We never pass a lookup on to an AI provider, and the model runs at the provider of the AI tool, under your agreement with that provider. Where do AI providers process the data? shows what the most common providers state about location.

What personal data can an AI tool get from Placepoint?​

Owners of property who are private individuals are shown as Private person, and national identity numbers are never delivered, whatever access your account has otherwise. The extended access some customers have through a documented legitimate interest applies in Fusion, not in Placepoint AI. This is data minimisation under GDPR Article 5(1)(c) and data protection by design under Article 25. We chose it because the data ends up in a tool we do not control. The personal data that remains is already public: names of role holders in companies from Brønnøysundregistrene, that is the general manager and the board, and names and years of birth of shareholders from Aksjonærregisteret (the shareholder register), which Skatteetaten (the Norwegian Tax Administration) publishes. Shareholders are shown as Private person until the AI tool asks for names. Names of role holders and shareholders are delivered whether or not the account is approved for personal data, so here an AI tool can get names that Fusion hides for the same account. National identity numbers are never delivered. Details: Personal data is not available.

What does Placepoint log when an AI tool fetches data?​

We log every lookup from Placepoint AI with the tool that was called, what the AI tool sent in the lookup, and who asked: including user ID, IP address and organisation. We store the content of the lookup, such as search terms, addresses and matrikkel numbers, truncated after 2,000 characters. We do not log the answer we send back, only its size. The log is stored in Microsoft Azure in Norway East and is deleted after 90 days. This applies to Placepoint MCP and Placepoint Dataset MCP. Lookups to Placepoint AI for Sweden are processed at AWS in Ireland.

The log has two purposes. The first is operations and troubleshooting. For this, you are the controller for the content of the lookup and we are the processor, in the same way as for a search in Fusion. The second is to be able to trace who has fetched what, and we do this as controller.

The difference from Fusion is that the AI tool writes the lookup for you. If the tool has access to your own documents, the model can include text from them in a lookup. You control this in the AI tool, with instructions about what lookups should contain. You can find a ready-made example in Send only what the lookup needs.

We can also turn on minimal logging for your organisation. We then log only which tool was used, who asked and when. We do not store the content of the lookup, the IP address or error messages. This applies to all users in your organisation. Contact your contact person or support@placepoint.no to turn it on.

What about datasets I upload through Dataset MCP?​

They are handled the same way as when you upload them in Fusion: they are stored with us in Azure Norway East, are visible only to users you have given access to the project, and are covered by the data processing agreement. You decide what you upload and whether it contains personal data. If it contains personal data, the data processing agreement must be in place first.

Does Placepoint use the data to train models?​

No. We do not train any models on customer data or usage data, and we do not share it with third parties beyond the operations suppliers in the list of sub-processors. Anthropic, which supplies the language models behind the Agent, map layer chat and filter chat in Fusion, may not train models on what we send there either. This is set out in Anthropic's commercial terms. If the AI tool you have chosen uses data for training, that is governed by your agreement with that provider, not by us.

Does the same apply to the Agent in Fusion?​

The Agent runs in Fusion with your login and follows your access level in the platform, as the Agent page describes. It is in pilot and requires a licence: we enable it for the users your organisation has bought licences for. The Agent uses language models from Anthropic in the USA. So do map layer chat and filter chat in Fusion, which are on for all users and send the text the user writes in them. Anthropic is in the list of sub-processors.

Names and years of birth of private individuals are not sent to Anthropic, see the next question.

Does the language model behind the Agent get names of private individuals?​

Not the names of property owners. When the Agent looks up who owns a property, the language model gets a reference instead of the name. Fusion inserts the name only when the answer is shown to you, with the same access control as in the panels:

  • If your organisation is approved for personal data, you see the name, as you do in the Owners tab.
  • If it is not approved, you see Private person.

The language model gets neither names, years of birth nor national identity numbers of private individuals from the matrikkel or grunnboken, whatever access you have. This is data minimisation under GDPR Article 5(1)(c), chosen because the model runs at a provider in the USA.

Information from open registers is the exception. Board members, the general manager and the auditor are publicly listed at Brønnøysundregistrene, and shareholders are listed in the shareholder register, which is public. If you ask about roles or owners in a company, the language model can get the names and years of birth of these people. It never gets national identity numbers. Companies are shown with their name and organisation number.

This leads to one reservation for property owners. If a private individual who owns property also holds a position or shares in a company, their name is in the open registers. If both the property and the company come up in the same conversation, the language model can link the reference to the name.

Two things can still reach the language model: a name you type in the question yourself, and content from systems you have connected to the Agent, such as documents and datasets. That is your own data, and you choose what you send.

DORA and financial institutions​

Does DORA apply to Placepoint?​

Not directly. DORA is the EU regulation on digital operational resilience in the financial sector, and it applies in Norway through the DORA Act from 01/07/2025. The requirements apply to banks, insurance companies and other financial institutions supervised by Finanstilsynet (the Financial Supervisory Authority of Norway). Only suppliers designated as critical at EU level are under direct oversight, and Placepoint has not been designated.

For a financial institution, we are an ICT third-party service provider, because Fusion, the Placepoint API and Placepoint AI are digital and data services provided on an ongoing basis. The agreement with us normally belongs in your register of information under Article 28(3), and it must include the terms in DORA Article 30.

Does Placepoint accept the contractual terms in DORA Article 30?​

Yes. Get in touch with your contact person or support@placepoint.no, and we will agree the terms in Article 30(2) with you. If your organisation has its own template for DORA terms, we are happy to use it.

You assess whether the service supports a critical or important function. If it does, we also agree the terms in Article 30(3), which cover service levels, audit and a plan for moving to another provider, among other things.

What does it mean that the service is not critical or important?​

DORA still applies to the agreement, but with fewer requirements. You do not need the terms in Article 30(3), nor threat-led penetration testing (TLPT) under Article 26 or the other requirements that apply only to providers of critical or important functions. The agreement must still include the terms in Article 30(2) and be in your register of information.

Storage and operations shows where the data is stored, and the list of sub-processors shows who processes it for us.

For procurement and data protection officers​

What can I ask Placepoint for?​

How do you fill in our security questionnaire?​

We answer the parts that apply to our delivery, and mark the rest “not relevant” with a short justification, for example requirements that apply only to providers of critical functions under DORA. The answers describe the level we have when the agreement is signed. Where we do not fully meet a requirement, we answer no or partly. Short summary for supplier questionnaires shows what we write most often.

Who is the contact point for privacy?​

support@placepoint.no. Your enquiry goes to the right person at Placepoint, and we answer requests for access, rectification and erasure within 30 days, as the privacy policy describes.

I am listed in grunnboken or the company register. What does Placepoint hold about me?​

What the sources have registered: for a title holder, name, year of birth and ownership share from grunnboken; for a role holder, name and role from Brønnøysundregistrene. We do not collect more about private individuals than what is in the sources, and we show private individuals as Private person to customers who have not documented a need. If you want to know what is registered about you, or think something is wrong, contact us at support@placepoint.no. Errors in the source are corrected by Kartverket or Brønnøysundregistrene (BRREG), and the correction reaches us at the next update.

Sub-processors​

These are the suppliers we use to operate Placepoint, which therefore process personal data on our behalf. The data processing agreement refers to this list, so we keep it up to date here.

Before we start using a new sub-processor, we update the list and notify customers with a data processing agreement by email at least 30 days in advance. You can object within that period.

Last updated 20/11/2026.

VendorServiceProcessesCountry and regionTransfer mechanism
Microsoft Ireland Operations Ltd. (Azure)Hosting, databases, storage, login (Entra)All data in the platform, including your own datasets and the user databaseNorway, Norway East with a replica in Norway WestEEA
Amazon Web Services (AWS)Hosting of Placepoint AI for Sweden, the server for Swedish propertyLookups from AI tools about Swedish property, with the user ID from the loginEU, IrelandEEA
PostHog Inc. (EU service)Product analytics and feature management in FusionUser ID, events in the interfaceEU, GermanyEEA
Mixpanel Inc. (EU service)Usage analyticsUser ID, events in the interfaceEUEEA
Better Stack GmbHOperations log for the servicesUser ID, tenant ID, requests to the servicesEU, GermanyEEA
Twilio SendGrid (EU service)System email, such as invitations and notificationsName, email address, content of the notificationEU, GermanyEEA
ipapi.isLookup of country and city for the IP address at loginThe IP address, truncated to the network it belongs toEU, GermanyEEA
Functional Software Inc. (Sentry)Error loggingUser ID, email address, IP address, error messagesUSAEU-US Data Privacy Framework
HubSpot Inc.Customer system, support chat and emailName, email address, phone number, enquiriesUSAEU-US Data Privacy Framework
Slack Technologies LLCMirroring of support chat to the support teamThe content of chat enquiriesEUStandard contractual clauses
Anthropic, PBC (Anthropic Ireland, Limited as contracting party for the EEA)Language models for the Agent, map layer chat, filter chat and 3D model generation in FusionThe text the user writes in these features. For the Agent, also the conversation and lookup results. Private individuals who own property are replaced with a reference. Names from open registers (roles at Brønnøysundregistrene and the shareholder register) may be includedUSAStandard contractual clauses

What is not in the list​

Our data suppliers, such as Kartverket, Brønnøysundregistrene (BRREG) and the others on the supplier pages, are sources we fetch data from, not processors for us. They do not process any information about users or your datasets.

Map and analysis services, such as travel time and background maps, receive the coordinates and addresses in the lookup, not who asked.

You choose the AI tool you connect Placepoint AI to, and it is covered by your agreement with that provider. For the reason why: Placepoint AI and your own datasets.

Short summary for supplier questionnaires​

These are the points procurement and IT security ask about, with a link to the full answer. If you need your questionnaire filled in, send it to support@placepoint.no.

  • Hosting and storage: Microsoft Azure in Norway East, with a replica in Norway West. Placepoint AI for Sweden runs at AWS in Ireland. Where is the data stored?
  • CLOUD Act: Microsoft and most of the sub-processors are American companies and are covered by the law. We have no measures of our own beyond Microsoft's commitments in its data processing agreement. Does US law, such as the CLOUD Act, apply to the data?
  • Login: a personal account for each user, email and password, and a one-time code by email when the account is activated or the password is changed. Account and login
  • Use outside the EU/EEA: no technical block, but every login is logged with its IP address, and the terms of use require use from the EU/EEA. Access outside the EU/EEA
  • Encryption: TLS for all traffic, encryption at rest in the databases. How is the data secured?
  • Access control: personal data about private individuals is opened per account after a documented need, and access to production data is limited to named operations staff. Access to personal data
  • Traceability: the usage log shows who has seen which personal data. What personal data does Placepoint process about users?
  • Placepoint AI: no language model, no data to an AI provider. Owners of property are shown as Private person and national identity numbers are never delivered, while names of role holders and shareholders can be delivered. Lookups to the Norwegian servers are logged in Norway East and deleted after 90 days, and minimal logging can be turned on per organisation. Placepoint AI and your own datasets
  • AI in Fusion: the Agent (licence per user), map layer chat and filter chat (on for everyone) use language models from Anthropic in the USA under standard contractual clauses. The Agent does not send names or years of birth of private individuals who own property to the model. Names from open registers (roles and shareholders) may be included, also for an owner who is listed there. Does the same apply to the Agent in Fusion?
  • Training: no models are trained on customer data or usage data, including at Anthropic. Does Placepoint use the data to train models?
  • Incidents: written notice within 48 hours for breaches affecting your data. What happens in a security breach?
  • Agreements: data processing agreement under Article 28, our template or yours, and terms under DORA Article 30. If the service is not critical or important for you, only Article 30(2) applies. Data processing agreement, DORA and financial institutions
  • Sub-processors: published list with service, country and transfer mechanism. Sentry and HubSpot are in the USA under the EU-US Data Privacy Framework, Anthropic is in the USA under standard contractual clauses, and the rest are in the EU/EEA. New sub-processors are announced at least 30 days in advance. Sub-processors
  • Deletion: your own data is deleted or returned when the agreement ends, confirmed in writing. Backups expire on a fixed schedule, the annual ones after ten years, and operations and application logs are deleted after 30 to 90 days. How long does Placepoint store the data?

Where do AI providers process the data?​

Our part is the same whatever the tool: our servers run in Norway, Placepoint AI for Sweden runs in Ireland, and we send nothing to an AI provider. Where the model runs and what is stored there depends on your agreement with the provider of the AI tool. The table shows what the most common providers state themselves, checked on 30/10/2026, and Microsoft Foundry on 31/10/2026. Ask your provider to confirm what applies to your agreement before you draw conclusions.

AI toolWhat the provider statesSource
Claude (claude.ai and Anthropic's API)Storage in the USA. Traffic can be processed in the USA, Europe, Asia and Australia, and you can choose processing in the USA only. No EU region.Anthropic: Where are your servers located?
Claude in Amazon Bedrock or Google Vertex AIThe Claude models can run in EU regions, with AWS or Google as processor instead of Anthropic. Requires a separate setup and an AI tool that supports it.Claude in Amazon Bedrock, Claude Opus 5.5 on Google Cloud
Claude in Microsoft FoundryTwo options. Hosted on Azure runs the model on Azure, stores the data in the Azure geography you choose and processes it globally or in the USA. Foundry has no EU zone for Claude. Hosted on Anthropic can process the data outside Azure and outside the region. Anthropic is the processor in both.Microsoft: Compare hosting options for Claude models in Microsoft Foundry
ChatGPT Enterprise and EduYou can choose storage in Europe (EEA and Switzerland), and eligible customers can also have the model run in Europe. Login, routing and other processing that does not use GPUs can take place outside the region.OpenAI: Data residency and inference residency for ChatGPT
Microsoft 365 CopilotAn EU Data Boundary service for EU customers, so traffic stays within the EU boundary. Models from Anthropic in Copilot are excluded. Prompts and responses are not used to train models.Microsoft: Data, Privacy, and Security for Microsoft Copilot
Models hosted in the EU, such as Mistral, Azure OpenAI in an EU region or self-hostingThe location follows the provider or hosting you choose. Placepoint AI works with all MCP clients, see Other MCP clients.The provider's own documentation

Common questions about AI and data storage​

Short answers to the questions we most often get from procurement, data protection officers and IT security, whether you use Claude, ChatGPT, Copilot or another AI tool. Each answer links to the full answer.

Does Placepoint store or process our data in the USA?​

The platform, the databases, the map layers, the datasets you upload and Placepoint AI run in Microsoft Azure in Norway. The exception is Placepoint AI for Sweden, which runs at AWS in Ireland. Three things go to the USA: error logging and the customer system (Sentry and HubSpot), the language models behind the Agent, map layer chat and filter chat in Fusion (Anthropic), and address search, which uses Azure Maps on a global endpoint. Microsoft and several of the other suppliers are also American companies, so US law applies wherever the data is stored. See Where is the data stored?, CLOUD Act and Sub-processors.

Does Placepoint AI pass our lookups on to an AI provider?​

No. The servers contain no language model and only answer lookups from the AI tool you have connected yourself with your own login. See Who is the controller when an AI tool fetches data from Placepoint?.

We use Claude, ChatGPT or Copilot. Where does the data end up?​

With the provider of the AI tool, under your agreement with that provider. Where do AI providers process the data? shows what the most common providers state about location.

Can we keep all processing within the EU/EEA?​

For Placepoint AI, almost. The servers run in Norway, Placepoint AI for Sweden runs in Ireland, and with an AI tool that runs the model in the EU/EEA, such as Claude through Amazon Bedrock in an EU region, ChatGPT Enterprise with processing in Europe, or a model you host yourselves, the model also runs in Europe. Claude through Microsoft Foundry cannot be kept in the EU/EEA, see the table of AI providers. The exception on our side is address search through Azure Maps, which uses a global endpoint. The Agent, map layer chat and filter chat in Fusion use Anthropic in the USA and cannot be kept in the EU/EEA today. Contact us, and we will go through the setup with you.

Does the AI tool see names and national identity numbers of private individuals?​

Partly. Owners of property are shown as Private person, and national identity numbers are never delivered. Names of role holders in companies, and names and years of birth of shareholders, can be delivered when the AI tool asks for them, whether or not the account is approved for personal data in Fusion, because they are public at Brønnøysundregistrene and Skatteetaten. See What personal data can an AI tool get from Placepoint?.

What do you log about lookups from the AI tool, and for how long?​

The tool that was called, what the AI tool sent, truncated after 2,000 characters, and who asked. For the Norwegian servers, the log is stored in Norway and deleted after 90 days, and the answer we sent is not logged. See What does Placepoint log when an AI tool fetches data?.

Can text from our own documents end up in your log?​

Yes, if the model includes text from the documents in a lookup. You control this with an instruction in the AI tool, and with minimal logging we do not store the content of the lookup at all. See Send only what the lookup needs and What does Placepoint log when an AI tool fetches data?.

Do you train models on our data?​

No. We do not train any models on customer data or usage data, and Anthropic may not train on what the AI features in Fusion send there. Whether the AI tool you use does so is governed by your agreement with that provider. See Does Placepoint use the data to train models?.

What about the Agent in Fusion?​

The Agent runs in Fusion with your login and follows your access level. It is off until your organisation has bought licences for the users who should have it. It uses language models from Anthropic in the USA, which is in the list of sub-processors. Names of private individuals who own property are not sent there, with one reservation for owners who are also listed in open registers, see Does the language model behind the Agent get names of private individuals?.

Can we turn off AI for our users?​

Mostly. Without a seat on the licence for Placepoint AI, no one in your organisation can connect an AI tool to Placepoint, and the Agent is off until you buy a licence. Map layer chat and filter chat in Fusion are on for all users and have no switch per organisation today. They send only the text the user writes in them. See Who has AI access?.

Do you sign a data processing agreement, and can we use our own template?​

Yes and yes. We have a template reviewed by lawyers with four appendices, and we are happy to fill in yours, with two fixed points about register data and the list of sub-processors. See Data processing agreement.

Does DORA apply, and do you accept the terms in Article 30?​

DORA applies to the financial institution, and for you we are an ICT third-party service provider. We agree the terms in Article 30(2), and in Article 30(3) if the service supports a critical or important function. See DORA and financial institutions.

Who are your sub-processors, and which of them are in the USA?​

The list is on this page with service, country and transfer mechanism. Sentry (error logging) and HubSpot (customer system) are in the USA under the EU-US Data Privacy Framework, Anthropic (language models for the AI features in Fusion) is in the USA under standard contractual clauses, and the rest are in the EU/EEA. See Sub-processors.

How is login secured?​

Each person has their own account and logs in with email and password, and a one-time code by email confirms their identity when the account is activated or the password is changed. Every login is logged with its IP address, and the terms require use from the EU/EEA. See Account and login.

Who answers questions about privacy?​

support@placepoint.no. We answer requests for access, rectification and erasure within 30 days. See Who is the contact point for privacy?.

See also: Integration and security, Account and login, Data and sources, Map and properties, Placepoint AI, Data processing agreement (glossary), GDPR (glossary).

These pages are generated automatically and may contain errors.This applies to screenshots and videos in particular. See About these help pages. Let us know through "Did you find what you were looking for?" at the bottom of the page, the chat at bottom right or support@placepoint.no.