Building Ethical AI at Modzy

Reading Time: 3 minutes

Modzy is leading by example – committed to the idea that together, we can build AI free of bias, discrimination and that upholds a set of ethics principles. We all have important roles to play in building ethical AI.

This blog posts shares background on how we developed our Community Guidelines.  Ethics and transparency are critical to building trust between people and ever-growing machine learning systems making automated decisions.

Modzy Community Guidelines



Almost. More or less. Good try. After spending years observing missteps and consuming hundreds of articles that merely paid lip service to the idea of ethical AI, the leadership at Modzy knew we needed to put our beliefs and thoughts into REAL action. We focused on a bottoms-up approach that prioritized AI built with transparency, technical robustness and safety, privacy and data governance, diversity, non-discrimination and fairness, and accountability. We put ethics first so our customers could achieve AI success the right way.

Our leadership team engaged in spirited debates about what types of AI models we would or wouldn’t allow in the marketplace, and our data science team established processes for building models that allowed for multiple checkpoints to ensure that they were being built to behave in an ethical manner. At the end of the day, however, we didn’t think this went far enough, and we wanted to ensure that we weren’t just abiding by principles that we thought were important when it came to building trusted AI. We wanted to ensure that our thinking reflected society’s evolving opinion on how to address the issue of AI ethics.

In line with this vision, the Modzy product team analyzed over 100 international standards documents published by committees, research and academic institutions—each provided a point-of-view or set of recommendations on how best to wield and build AI in an ethical way [1]. Several guidelines stood out as best practices, and we also felt it was important  to derive our standards from previously developed standards by experts from a broad variety of disciplines.

Conducting this exercise was illuminating, and we realized that we were already thinking about many of the same things the guidelines called out for building and ensuring that our AI remains trustworthy and ultimately, Principled.

Our ethics guidelines are manifest in the actions and responsibilities of three key groups: 1) our leadership team, 2) our data science team, and 3) our partner community.

Our ethics principles span several foundational areas:


Transparency is at the heart of what we offer, which you can see in our commitment to sharing our guiding ethical principles, our community standards that show the types of companies and models we want (or don’t) in our marketplace, all the way down to the information we share about model performance.

Technical Robustness and safety

AI systems should be resilient against attempts to mislead them or against other types of attacks, which is why security is at the heart of Modzy, with embedded governance, enterprise-grade authentication, API security, and our patent-pending adversarial defense solutions.

Privacy and data governance

Trustworthy AI must align to and protect our values, including the right to privacy. Modzy is built to comply with the most stringent logging, auditing, and governance requirements, to ensure that data and access remain protected.

Diversity, non-discrimination, and fairness

AI technology must be built in service of a society that strives for fair equitable treatment of all people, which is why we ensure that diverse stakeholders are involved in every part of the Modzy build process, and why we won’t allow certain types of models into our marketplace; we’re also transparent about any biases found in model performance to minimize any impact to fair and equitable treatment once models are deployed.


Like any other IT system, AI must be accountable to the people overseeing it. Modzy contains numerous features, from embedded governance, user access control, logging and auditability, explainability, and model management and monitoring to ensure that you can always have insight into your AI performance.

This commitment to building AI that adheres to these principles is manifest not only in our leadership team, but also our data science team and partner community.

Data Science Team

Building ethical AI needs to be shared across all parts of an organization. Our data science and development teams carry the responsibility for creating trustworthy AI. They leverage a checklist for the model build process to ensure that they’re evaluating the right characteristics and that ethics considerations aren’t an afterthought.

Partner Community Guidelines

Ensuring that ethical AI remains a fundamental part of the Modzy foundation also means building trust across our partner community and sharing a collective vision for building and creating the best AI. To that end, our partner evaluation process includes steps that evaluate potential partners for their alignment with our thoughts on building ethical AI. By becoming a member of the Modzy marketplace, these companies also agree to adhere to our community guidelines, which stress the importance of a values-driven approach and commitment to build  ethical AI.

Ethics shouldn’t be an afterthought, or a roadblock, for building trusted AI. That’s why Modzy has prioritized ethical development at all levels of our organization. We hope to set a high standard for everyone operating in our industry.

[1] Jobin, Anna., Ienca, Marcello, and Vayena, Effy. The global landscape of AI ethics guidelines. Nat Mach Intell 1, 389–399 (2019).

Share this post
Share on facebook
Share on twitter
Share on linkedin
Share on email
Recent Updates

ModelOps vs. MLOps

Decisions organizations make today about building their AI pipelines will have major impacts on the future of AI maintenance for their organizations. But is everyone

Read More »