Skip to content
NewISO 9001:2026 Internal Auditor Course: enrollment open, places still available
International Certification Services

Home › Blog › ISO 27001 and AI

Using generative AI at work? How ISO/IEC 27001 helps you control the risks

AI assistants save time, but they can also leak contracts, client data and source code. An information security management system gives you rules that hold up in an audit.

Using generative AI safely with ISO/IEC 27001

6 min readBy the ICS technical team

Generative AI tools went from curiosity to daily habit in a very short time. In many companies, people already use them to draft emails, summarize contracts, write code or analyze spreadsheets, often with personal accounts and without anyone deciding what information can go into them.

ISO/IEC 27001 does not mention any AI tool by name, and it does not need to. Its method, based on assessing risks and choosing controls, applies to AI the same way it applies to email, the cloud or a USB drive.

At a glance

  • Treat AI tools as information processing services: assess the risk before people use them.
  • Decide which information can and cannot be entered into each tool.
  • Review AI providers like any other supplier that handles your data.
  • Train people and keep a record that you did.

The real risks

  • Data leakage. Confidential information typed into a tool may be stored or used by the provider, depending on the service and its settings.
  • Personal data. Client and employee data entered into an AI tool is still personal data under Panama’s Law 81 of 2019.
  • Wrong answers. AI output can be inaccurate, and decisions based on it need human review.
  • Shadow AI. Tools adopted by staff without approval, outside any control.

The ISO/IEC 27001 controls that apply

Several controls in Annex A of ISO/IEC 27001:2022 address these risks directly:

  • 5.10 Acceptable use of information and other associated assets: what staff may and may not enter into AI tools.
  • 5.19 to 5.23, supplier relationships and cloud services: evaluating AI providers, their terms and where data is processed.
  • 5.34 Privacy and protection of PII: keeping personal data out of tools that are not approved for it.
  • 6.3 Information security awareness, education and training: teaching the rules with real examples.
  • 8.12 Data leakage prevention: technical measures where the risk justifies them.

A simple AI policy in five steps

  1. List the AI tools people already use, including free ones.
  2. Classify your information and decide which classes may be used with which tools.
  3. Approve a short list of tools, preferably business versions with clear data terms.
  4. Write a one-page acceptable use rule and train everyone on it.
  5. Review incidents and the approved tools at your management review.

What about ISO/IEC 42001?

ISO/IEC 42001, published in 2023, is a management system standard specifically for artificial intelligence. It is most relevant for organizations that develop or provide AI-based products and services. For most companies that simply use AI tools, a well-implemented ISO/IEC 27001 system is the practical starting point.

How ICS can help

We can include AI use in the scope of a gap analysis or pre-audit of your information security system, and certify your ISO/IEC 27001 management system when it is ready.

Share this article

Bringing AI into your information security system?

We review your controls in a gap analysis and certify your ISO/IEC 27001 system when you are ready.

More articles

Ready to start your certification process?

Tell us which standard you need, the size of your operation and which country you are in. We will reply with a clear proposal.