Bentley AI User Guidelines

Last updated: August 01, 2026

 

These AI User Guidelines (“Guidelines”) govern Subscriber’s use of Bentley’s AI products and features (collectively, “AI Products”). In addition, where a Subscribers agreement with Bentley is governed by Bentley’s commercial program agreement, then the use of AI Products is also governed by the AI Product Specific Terms, which are incorporated by reference into these Guidelines.

  1. Co-Innovation Initiative

Bentley operates co-innovation initiative to better understand how Subscribers are integrating our AI Products into applications and workflows. As AI transforms infrastructure workflows, Subscribers are invited to share AI-driven use cases and partner with Bentley to leverage AI Products in novel and creative ways. Contact us at Infrastructure.AI@bentley.com.

  1. Disallowed Usage

Subscribers may not use AI Products, nor allow their users or any third party to use AI Products, for the following:

  • Automated Decision-Making Processes with Legal Effects. As part of an automated decision-making process with legal or similarly significant effects, unless Subscriber ensures that the final decision is made by a human being. In this case, Subscriber must take account of other factors beyond the AI Products’ recommendations in making the final decision.
  • Individualized Advice from Licensed Professionals. Relying upon individualized advice, in the ordinary course of business would be provided by a licensed professional. Bentley’s AI Products are not substitutes for the Subscriber’s professional judgment.
  • Explicitly Predicting Protected Characteristics. Explicitly predicting an individual’s protected characteristic, including, but not limited to, racial or ethnic origin, and past, current, or future political opinions, religious or philosophical beliefs, trade union membership, age, gender, sex life, sexual orientation, disability, health status, medical condition, financial status, criminal convictions, or likelihood to engage in criminal acts.
  • Weapons Development. Developing, advertising, marketing, distributing, or selling weapons, weapon accessories, or explosives, as enumerated by the United States Munitions List.
  • Subscribers may not deceive end users or consumers by misrepresenting content generated through automated means as human generated or original content.
  1. Model Context Protocol (MCP) Usage

MCP enables third-party AI Clients to connect to and act upon Bentley systems, tools, and data through MCP Servers. Subscriber must treat the grant of MCP access to an AI Client in the same manner as granting a person, application, or integration access to Bentley systems, including applying the same governance, approval, and risk-assessment standards.

  • Approved AI Clients Only. Subscriber must only connect approved AI Clients to MCP Servers. Subscriber is responsible for ensuring that any AI Client used is approved under Subscriber’s enterprise AI governance program and is approved for the data classification of the information the AI Client may access.
  • Data Classification Controls. Subscriber must not connect an MCP Server to an AI Client were doing so could expose customer data, personal data, confidential information, export-controlled information, regulated infrastructure information, or sensitive engineering models, unless such connection is expressly permitted under Subscriber’s data governance policy and subject to approved technical and organizational controls.
  • Review of AI-Initiated Actions. In addition to reviewing AI-generated content, Subscriber must review all tool calls, API requests, model modifications, engineering workflows, generated scripts, and proposed changes initiated by an AI Client via an MCP Server before execution or reliance.
  • Authentication and Credential Handling. Subscriber must not share access tokens or credentials with, or on behalf of, an AI Client; hard-code credentials into any MCP configuration; bypass enterprise authentication controls; or grant an AI Client permissions beyond those necessary for its intended function. Subscriber must apply least-privilege access principles and rely only on approved authentication mechanisms when connecting AI Clients to MCP Servers.
  • MCP Registry and Connector Governance. Subscriber must only use MCP Servers and MCP registries approved under its enterprise AI governance program. Open-source or third-party MCP integrations must undergo security review prior to enterprise adoption.
  1. Notices

AI technology, including generative AI, will continue to be used in new and innovative ways. Subscriber is responsible for determining if its use of these technologies is safe and in accordance with applicable laws.

  1. Disclaimer

Subscriber’s use of Bentley’s AI Products will result in outputs including but not limited to recommendations and/or suggestions (collectively, “Output”). Output from AI Products is generated through machine learning and artificial intelligence. Bentley has not verified the accuracy of the Output, and it does not represent Bentley’s views. Bentley makes no warranty or guarantees as to the accuracy, completeness, or reliability of the Output and does not accept any liability or responsibly arising in any way from Subscriber’s use of the Output or any omissions or errors contained in the Output.

  1. Feedback

We welcome feedback on our AI Products. You can use the report button, get in touch with our Support team, and/or contact ai@bentley.com to provide feedback at any time.