What Goes Into an SAP AI Ethics Review? Our Experience with docuflow AI for SAP Solutions

A behind-the-scenes look at SAP’s AI Ethics Review process and our experience with docuflow AI for SAP, from assessing potential risks and documenting safeguards to demonstrating responsible AI governance, transparency, and human oversight.
What goes into an SAP AI ethics review: the SAP AI ethics principles reviewed for docuflow AI for SAP, which are fairness, privacy, transparency, human oversight, and AI-generated content.

On this page

As AI becomes more integrated into enterprise business processes, ensuring that it operates responsibly is just as important as delivering accurate answers. As part of our SAP Integration and Certification Center (ICC) certification process, we recently completed an AI Ethics Review for docuflow AI for SAP Solutions. The experience gave us a closer look at how SAP evaluates AI applications, the safeguards expected, and the documentation needed to demonstrate responsible AI practices. 

Understanding SAP's AI ethics review process

The process began with two assessments: SAP’s AI Business Scenario Impact Self-Assessment and AI Ethics Policy Self-Assessment, supported by an AI Ethics Assessment Checklist. The first assessment focused on understanding the application’s purpose, how it uses AI, and the potential risks associated with its intended business scenarios.

  • Does the application process personal information, and how is that information protected? 
  • Could the AI influence decisions that have consequences for individuals? 
  • Is the application involved in automated decision-making? 
  • Could the application be used for surveillance, discrimination, manipulation, or other prohibited activities? 
  • What additional safeguards are in place for high-risk industries or business processes? 

One interesting outcome was that our solution was pre-classified as high-risk because it can support use cases in areas such as critical infrastructure and Human Resources. Although docuflow AI for SAP Solutions provides informational assistance rather than making autonomous decisions, SAP’s assessment recognizes that even advisory AI outputs can have consequences when used in sensitive business processes. 

Demonstrating AI safety and governance

The second assessment examined how we address SAP’s AI ethics principles through our application design, security controls, and operating processes. This meant documenting how we handle fairness, privacy, transparency, human oversight, and AI-generated content. 

During the review, we walked through the AI capabilities of docuflow AI for SAP Solutions and the safeguards available within the application. Our approach is built on SAP’s safeguards, extended by docuflow’s retrieval design. We leverage SAP AI Core Orchestration capabilities, including content filtering, data masking, Prompt Shield, and Llama Guard. Through our administration application, these safeguards can be configured at the SAP business-object level, allowing customers to apply different controls depending on the business scenario. We complement these capabilities with additional measures, including a scanner for prompt injections embedded in retrieved content, SAP-context-aware document retrieval, and our Prompt Inspector for reviewing the information and configuration behind AI responses. This flexibility is important because a Plant Maintenance scenario involving technical procedures may require different protections than an HR scenario involving employee policies. 

The documentation behind the review

A significant part of the process involved showing not just that safeguards existed, but how they were implemented, maintained, and tested. Our supporting documentation included the completed SAP assessments and checklist, along with our Technical Product Profile, Security & Data Privacy whitepaper, Responsible AI questionnaire, application administration guides, and technical documentation. Together, these materials addressed the solution’s intended use, architecture, data flows, access permissions, model usage, configuration options, known limitations, and governance processes. 

SAP’s assessment also examined testing and operational readiness, including documented test cases and results, handling of unexpected AI behaviour, acceptable error thresholds, monitoring, and mechanisms for gathering user feedback. Importantly, the process required us to identify areas still being formalized. Our self-assessment documented follow-up work involving structured adversarial testing, measurable acceptance criteria, monitoring, and other governance improvements. This is an important part of responsible AI: understanding not just which safeguards are available, but where additional verification or controls are needed. 

SAP AI ethics review outcome

Following our demonstration and review discussions, docuflow AI for SAP Solutions successfully passed SAP’s AI Ethics Review. SAP’s ethics reviewers confirmed that there were no further questions from an ethics perspective, allowing us to proceed with the next steps of the broader certification process. This outcome reflects the importance of demonstrating not only that responsible AI safeguards are in place, but also how they are documented, configured, and governed.

Moving forward with responsible AI

Following our demonstration and review discussions, SAP’s ethics reviewers indicated that there were no further questions from an ethics perspective, allowing us to proceed with the next steps of the broader certification process. For us, the experience reinforced that responsible AI involves much more than configuring a language model. It requires a clearly defined purpose, documented safeguards, controls that can be adapted to different business processes, and evidence that those controls operate as intended. Responsible AI isn’t simply about producing the right answer. It’s about understanding, controlling, and verifying the process behind that answer.

Our vision is for docuflow to become the trusted contextual skill connecting SAP AI and enterprise repository AI, enabling intelligent agents to work together using accurate, relevant business information. Whether an SAP agent such as Joule is leading a business process or a repository agent such as Box AI is managing a content-driven workflow, docuflow provides the missing context between them. It connects SAP business objects, transactions and process state with enterprise documents, metadata, permissions and authoritative sources, allowing either agent to understand not just what information exists, but why it matters to the business process at hand.

Through emerging interoperability standards such as MCP and A2A, we envision docuflow making these contextual capabilities available as reusable skills, independent of the customer’s choice of AI platform or content repository. Built on Clean Core principles, using supported SAP interfaces and side-by-side extensions rather than modifications to the SAP core, docuflow is designed to preserve these AI capabilities and business-context relationships through SAP upgrades, S/4HANA modernization and evolving AI architectures.

The result is a durable foundation for accurate, governed and increasingly autonomous enterprise AI—without requiring SAP agents to become content experts, repository agents to become SAP experts, or customers to rebuild their AI investments every time their technology landscape changes.

Key takeaways
  • SAP’s AI Ethics Review evaluates more than AI accuracy. The process examines potential business risks, privacy, fairness, transparency, human oversight, and responsible AI governance.

  • AI applications can be classified as high-risk even without autonomous decision-making. docuflow AI for SAP Solutions was pre-classified as high-risk because it can support sensitive business scenarios, including critical infrastructure and Human Resources.

  • AI safeguards must be configurable for different business scenarios. docuflow leverages SAP AI Core Orchestration capabilities alongside SAP-context-aware retrieval, prompt injection scanning, and the Prompt Inspector to support scenario-specific controls.

  • Documentation and testing are essential to responsible AI. SAP’s review examined technical documentation, security controls, application architecture, testing practices, monitoring, and governance processes.

  • Responsible AI requires continuous improvement. The review reinforced the importance of identifying areas for further verification, strengthening testing practices, and maintaining oversight as AI capabilities evolve.

docuflow AI for SAP Solutions

See how docuflow connects SAP processes and governed enterprise content to support context-aware AI use cases.

Frequently asked questions

SAP’s AI Ethics Review is an assessment process that evaluates the potential risks and responsible AI practices associated with an AI application. It examines areas such as fairness, privacy, transparency, human oversight, security, and the safeguards implemented to address potential risks.

docuflow leverages SAP AI Core Orchestration capabilities, including content filtering, data masking, Prompt Shield, and Llama Guard. Additional measures include prompt injection scanning for retrieved content, SAP-context-aware document retrieval, and the Prompt Inspector, which helps review the information and configuration behind AI responses.

docuflow connects SAP business context with enterprise documents, metadata, permissions, and authoritative information sources. This helps AI applications retrieve information relevant to specific business objects and processes while supporting governance and repository choice. Its longer-term vision includes enabling interoperable AI agents through emerging standards such as Model Context Protocol (MCP) and Agent2Agent (A2A).

 
Keep reading

Related posts

VersaFile Named a 2026 Deloitte Technology Fast 50 Companies-to-Watch Winner

VersaFile Inc. has been named a 2026 Deloitte Technology Fast 50™ Companies-to-Watch winner, recognizing its growth, innovation, and commitment to advancing enterprise content management and intelligent automation through docuflow®.

What’s New with docuflow for SAP SuccessFactors

docuflow for SuccessFactors just picked up three updates that change how HR teams file, find, and work with their documents day to day.

How to Manage SAP Documents Without Bloating Your Core System

This article explores why SAP content becomes a hidden modernization blocker—and how organizations can regain control without rewriting SAP or introducing another monolithic ECM.
Subscribe

Stay informed on SAP content and automation

Get the latest docuflow insights, technical articles, customer stories, and event updates delivered to your inbox. Practical information to help you navigate SAP content management and modernization.