iSearch

Can you use ChatGPT at work?

Many people already use the service to summarise a text or draft a message. Before it becomes routine, you need to know what happens to what is pasted in.

iSearch

Contents
  1. A personal account and a workplace account
  2. What may be entered
  3. A clear line in the organisation
  4. How it works in iSearch

The question often comes from someone who has already tried it. ChatGPT is easy to open, and a draft or a summary can be ready in a few minutes. What is simple for the individual can still become a question for the organisation, especially if personal data, client material or internal information is pasted in.

There is no general rule that bans the tool in a workplace. What decides whether the use is appropriate is the terms of the account being used, and the kind of information that is entered. That assessment is better made together, rather than left for each member of staff to work out alone.

A personal account and a workplace account

A personal ChatGPT account is covered by the provider’s terms for individual users in Europe. Those terms say that what you enter may be used to provide, maintain, develop and improve the service. Anyone who does not want their content used to train models can turn that off in the account settings. The setting is controlled by the account holder, and it is easy to miss in ordinary work.

If the account was created with a work email address, the organisation may later take it over. The provider’s terms describe how an account tied to an organisation’s email domain can be moved into the organisation’s workspace, and how an administrator can then see the content and control access.

A workplace account, such as ChatGPT Business or ChatGPT Enterprise, is covered by the provider’s terms for businesses. There, content from the workspace is not used to train models by default, unless the organisation has agreed to that. The same product name can therefore mean different handling of what you write, depending on which account is used.

The provider’s terms for individual users in Europe also have an addendum for when the service is used for work. Work use is not automatically ruled out by the terms. What still has to be examined is what is entered, and whether the processing of personal data meets the GDPR.

What may be entered

Even when the terms allow use at work, the GDPR applies as soon as personal data is processed. The Swedish Authority for Privacy Protection (IMY) has described how an organisation that uses generative AI in its work is as a rule the controller for that processing. If a member of staff pastes personal data into a publicly available service, the employer can still be responsible.

IMY has also noted that services designed for private individuals may lack a processor agreement that meets the GDPR’s requirements. Without such an agreement, it is difficult to use the service for personal data in a way that holds up.

Sensitive information, client material and texts that should stay inside the organisation belong in a tool that you have chosen and take responsibility for. An internal AI service and a considered approach to sensitive information make that line clearer than a ban that no one follows for long.

A clear line in the organisation

The practical answer is an AI policy that staff can follow. It should say which tools may be used, for what kind of task, and what must never be pasted in. A policy that only says public services are unsuitable, without offering an alternative for everyday work, is rarely followed for long.

If personal data will be processed, you need an agreement with the supplier and answers about where the data is handled, who can access it and how long it is kept. In some cases an impact assessment is also required before the processing starts, when it is likely to result in a high risk to individuals.

How it works in iSearch

iSearch is an internal AI assistant for everyday work. The whole chain can run in your own environment, the language model included, or we run it for you in Sweden. Answers are built from the material you have connected, with the source shown, and access follows the permissions you already have.

Read more about the difference between internal and public AI, about security, or book a walkthrough.

Sources: OpenAI, Europe Terms of UseOpenAI, Enterprise privacyOpenAI, Services AgreementIMY, GDPR in the use of generative AIIMY, Generative AI and controllershipGeneral Data Protection Regulation (EU) 2016/679

Related articles

All articles
  1. Must you negotiate with unions before introducing AI?

    When an AI tool changes how people work, Swedish co-determination law can require negotiations before the decision. What MBL says and what to discuss.

  2. How to write an AI policy your staff can follow

    How to write an AI policy that works day to day. Approved tools, which information may be used, responsibility, AI literacy and shadow AI.

  3. The cost of not using AI

    The hidden cost in many organisations is the time it takes to find the right information. How to make it visible, and how AI can give that time back.

Want to see how it would work for you?

Tell us what you want to solve, and we’ll show you how iSearch can help.