// ai & automation

Adding AI integrations into your work environment

Managing the integration of AI connectors with Microsoft 365 environments requires a clear boundary between the efficiency gains and the potential for data leakage or corruption.

Download the PDF

Risks of AI integration with M365 and SharePoint

The backup warning

Critical: Standard Microsoft 365 "Native Protection" is insufficient when deploying AI connectors. AI tools interact with data at scale and speed. In the event of an AI-driven "logic error" resulting in mass data corruption, over-writing of files, or accidental deletion, Microsoft's standard retention policies may not allow for a granular "undo" of the damage.

A third-party, point-in-time backup solution is mandatory to ensure data can be rolled back to a state prior to the AI integration or any specific AI-led incident.

AI Connector Implementation Disclaimer

  1. Scope of service. TranTech Computers ("the Company") provides the technical bridge between AI services and your Microsoft 365 environment. The Company does not control the logic, accuracy, or output generated by the third-party AI model.
  2. Data security and permissions. The Client acknowledges that the AI will have access to all data the authenticated user account can "see." It is the Client's responsibility to ensure that SharePoint permissions and folder access levels are strictly audited prior to implementation. The Company is not liable for the disclosure of sensitive information resulting from improper internal permission structures.
  3. Data integrity and backups. The Client understands that AI interactions can result in automated data modification. The Company strongly advises the use of an independent, third-party backup solution. The Company shall not be held liable for data loss, corruption, or "hallucinated" content produced by the AI connector.
  4. No guarantee of accuracy. Outputs generated by AI connectors are for informational purposes only. All AI-generated emails, summaries, and file modifications should be reviewed by a human professional.
  5. Acceptance of risk. By proceeding with this integration, the Client accepts all inherent risks associated with Large Language Models (LLMs) and third-party AI connectors, including potential impacts on data privacy and system stability.

To accept

Download the PDF, sign it and send it back to admin@trantech.com.au, or call us on (07) 4837 1201.

Name
Signed
Date

Next steps for implementation

Chat to us about AIBack to AI & Automation