AI guardrails: how is your data protected?

Julien Gruhier, head of AI considers the key issues of privacy and security, and how client data is protected when you upload a document in Croner Intelligence

Every AI supplier will tell you their product is secure. The more useful question should be more precise: once a client’s file leaves your computer screen, where does it go, who can read it, and how long does it stay there?

Who else sees what your employees ask the AI?

Most AI tools are not built by the company that built the underlying AI, such as Anthropic, OpenAI and so forth. As a result, the AI supplier sits in front of another company’s AI model, so always ask whether that other company sees your questions.

The AI tool can be set up so they never see the questions; that is the case with Croner Intelligence, which runs its models inside a sealed-off area of Amazon’s cloud that the AI model makers cannot enter on servers in the UK or the EU.

Your free features:

  • Breaking news and expert analysis
  • Customisable daily newsletters
  • Six free CPD learning modules each year
  • Personalised CPD tracker
  • Top 75 Firms league tables
  • Regulatory changes
  • Hardman’s Tax Data

Sign up to Business & Accountancy Daily

Related Articles
Subscribe