Toju Duke
AI is now a board-level decision, and most boards are making it without a defensible process. Legal teams flag risk, engineering teams ship models, and no one owns the question of whether the system should have been built at all. The gap between AI ambition and the controls needed to govern it is where reputational and regulatory damage accumulates.
Toju Duke is a former Google Responsible AI programme manager and author of Building Responsible AI Algorithms who helps organisations turn AI principles into operating controls.
Full Profile
Why organisations work with Toju Duke
- Ten years inside Google, including three running Responsible AI programmes across product and research teams on large-scale models. The framework she teaches is the one she ran.
- A published, structured Responsible AI framework covering fairness, transparency, safety, privacy and robustness across the full ML lifecycle, available to leadership teams as a shared reference point.
- Operator credibility on the question boards actually ask: how do we deploy AI at pace without inviting regulatory, reputational or human-rights exposure.
- Independent of any single vendor or platform. As founder of Bedrock AI and Diverse AI, she advises organisations on governance without selling them a model.
Biography highlights
- Author of Building Responsible AI Algorithms (Apress, 2023), a structured framework for fairness, transparency, safety, privacy and robustness across the machine learning lifecycle.
- Co-author with Prof Paolo Giudici of Responsible AI in Practice (Apress/Springer Nature).
- Ten years at Google. Final three years as Responsible AI Programme Manager working across product and research teams on large-scale models.
- Founder of Diverse AI, a community interest organisation for underrepresented groups in AI.
- Founder and CEO of Bedrock AI, an advisory and product company built around Responsible AI principles.
- Media contributor including BBC One Sunday Morning Live, Sky News, Al Jazeera, CGTN America, El Pais, La Vanguardia and Sifted.
Biography
Most organisations now have an AI policy. Very few have the operating mechanisms to make it real. Toju Duke spent the final three years of a ten-year career at Google building exactly those mechanisms, as a Responsible AI Programme Manager working across product and research teams on large-scale models.
That work became Building Responsible AI Algorithms, published by Apress in 2023. The book takes Responsible AI out of principles documents and into the machine learning lifecycle: how to define a problem, audit data, evaluate fairness, document a model, and decide whether to ship. A second volume, Responsible AI in Practice, co-authored with Prof Paolo Giudici, extends the framework into regulation, risk taxonomies and assessment.
Through Bedrock AI she advises organisations on governance and deployment. Through Diverse AI, a community interest organisation she founded, she runs education and research programmes for groups underrepresented in the field. The two operations are deliberate: governance and inclusion are the same question viewed from different sides of the model.
She speaks and writes for serious outlets, including BBC One Sunday Morning Live, Sky News, Al Jazeera, CGTN America, El Pais, La Vanguardia and Sifted. For a board weighing how to govern an AI programme it does not fully understand, she brings the rare combination of operator experience inside a frontier-model team and a published framework to organise the conversation around.
Key speaking topics
- Responsible AI frameworks and governance
- AI ethics and accountability
- Generative AI risk and oversight
- AI regulation and policy
- Fairness, bias and inclusion in AI systems
- Diversity in AI talent and design
- Implementing AI principles across the machine learning lifecycle
Ideal for
- Boards and executive committees commissioning or governing AI programmes
- Chief AI officers, CTOs, CIOs and chief data officers building Responsible AI operating models
- Risk, audit, legal and compliance leaders translating AI regulation into controls
- DEI and people leaders working on inclusion in technical organisations
Audience outcomes
- A working vocabulary for Responsible AI grounded in a published framework rather than vendor marketing.
- A view of where AI risk actually sits in the machine learning lifecycle, from problem definition through deployment.
- A reference model for translating AI principles into review processes, documentation and accountability structures.
- A sharper read on how regulation, fairness and human-rights considerations intersect for organisations deploying large-scale models.
Videos
Testimonials
Books
Fees
| EUR | GBP | USD | |
|---|---|---|---|
| Home Country | Under €12000 | Under £10,000 | Under $15000 |
| Asia Pacific | Under €12000 | Under £10,000 | Under $15000 |
| Europe | Under €12000 | Under £10,000 | Under $15000 |
| Middle East & Africa | Under €12000 | Under £10,000 | Under $15000 |
| South America | Under €12000 | Under £10,000 | Under $15000 |
| United Kingdom | Under €12000 | Under £10,000 | Under $15000 |
| US East Coast | Under €12000 | Under £10,000 | Under $15000 |
| US West Coast | Under €12000 | Under £10,000 | Under $15000 |
| Virtual | Under €12000 | Under £10,000 | Under $15000 |