--- title: Responsible AI Overview description: "Effective date: January 1, 2026 [Accept Mission](https://www.acceptmission.com/pt/encyclopedia/accept-mission/) is an innovation management platform for companies and innovation teams. We use AI-assisted features to strengthen workflows such as understanding ideas, improving descriptions, and generating structured summaries. We do not build our own…" url: "https://www.acceptmission.com/pt/knowledge/responsible-ai-overview/" updated: "2026-04-15" category: "AI Transparency & Responsible AI" --- # Responsible AI Overview Effective date: January 1, 2026 [Accept Mission](https://www.acceptmission.com/pt/encyclopedia/accept-mission/) is an innovation management platform for companies and innovation teams. We use AI-assisted features to strengthen workflows such as understanding ideas, improving descriptions, and generating structured summaries. We do not build our own large language model. We integrate with a pre-trained large language model through Microsoft Azure AI Foundry in a private environment. This page explains our AI principles, how AI is used in the platform, how we inform users, and how to report concerns. ## 1) Our AI principles We use AI to accelerate innovation workflows without reducing human control. Our principles are: ### Transparency We are clear about where AI is used and what it does. We aim to label outputs that are AI-generated or AI-assisted, and we explain what users should expect. ### Human oversight AI does not make decisions for customers. Users stay responsible for evaluation, approvals, and execution decisions. AI strengthens inputs and structure, but does not replace human judgment. ### Safety and misuse prevention AI features must not be used for harmful, illegal, or abusive activity. We apply usage rules and safeguards designed to reduce misuse, including limits and monitoring where appropriate. ### Privacy and data minimization We aim to keep AI requests focused on the task and avoid sending unnecessary personal data. Customers and users should not submit personal data or confidential information into AI features unless their organization has approved it and it is legally permitted. ### Accountability Customers control how AI features are used in their workspaces through policies, roles, and permissions. [Accept Mission](https://www.acceptmission.com/pt/encyclopedia/accept-mission/) maintains internal ownership for AI governance, platform safeguards, and issue handling. ## 2) Where users interact with AI in [Accept Mission](https://www.acceptmission.com/pt/encyclopedia/accept-mission/) AI-assisted features may appear in the platform when users: - Enrich an idea description or improve clarity of an idea - Generate structured summaries or key takeaways from content - Group or categorize similar ideas to support review - Support evaluation workflows with structured suggestions - Generate drafts that users can edit before saving AI features are designed to be optional at the point of use. Users choose whether to apply or ignore suggestions. ## 3) When outputs are AI-generated or AI-assisted AI can produce two types of output: ### AI-generated output Text that is generated by an AI model based on a user request, for example a summary or a draft description. ### AI-assisted output Text or structure that is suggested by AI but is intended to be reviewed, edited, and confirmed by the user, for example suggested tags, categories, or rewritten phrasing. Users remain responsible for verifying accuracy, completeness, and suitability before using AI output in decisions, reporting, or external communication. ## 4) How users are informed We inform users about AI in three ways: - In-product labeling: AI features are marked as AI or AI-assisted at the point where they are used. - Context guidance: prompts and helper text indicate what the feature does and what the user should review. - Governance pages: our Trust and Responsible AI pages explain the overall approach, limits, and reporting routes. If you notice an area where AI output is not clearly labeled, report it using the contact route below. ## 5) Model training and data use [Accept Mission](https://www.acceptmission.com/pt/encyclopedia/accept-mission/) does not train its own large language model. We use a pre-trained large language model provided through Microsoft Azure AI Foundry in a private environment. This setup is designed so that customer data is not used to train the underlying model. AI processing is used to return results to the platform for the specific request. Customers and users control what content is submitted for AI-assisted processing. We recommend avoiding personal data in AI prompts unless necessary and approved. ## 6) Safety, quality, and limitations AI output can be incorrect, incomplete, or misleading. It may also reflect bias present in training data or produce inconsistent results for similar inputs. Because of that: - Always review AI output before using it - Do not treat AI output as legal, medical, or financial advice - Validate critical claims using reliable sources and internal expertise - Use AI as a workflow accelerator, not a final authority ## 7) Staff training and oversight [Accept Mission](https://www.acceptmission.com/pt/encyclopedia/accept-mission/) staff who build, maintain, support, or oversee AI-assisted features receive internal guidance and training appropriate to their role. This includes: - Safe handling of customer data and access control expectations - Responsible use rules and misuse patterns - How to escalate AI concerns and potential incidents - Quality review practices for AI-assisted workflows