Video by Open Data Institute via YouTube
Harm assessment in impact assessments is notoriously difficult, often badly done, and long overdue for a rethink. In this Data Ethics Professional webinar, freelance privacy and data protection expert Rowenna Fielding launches the Data Hazards and Harms Toolkit — three years in the making, and built to replace guesswork with something you can actually use on Monday morning.
Most frameworks ask you to predict the unpredictable, then put a number on it. Rowenna’s argument is that data harms are foreseeable but genuinely unquantifiable, which is why the 5×5 matrix so often produces made-up numbers and false confidence. Instead, the toolkit borrows from health and safety: identify the hazardous materials, conditions and operations in your project, then layer safety measures around them — the way surgical teams prevent infection without hunting down individual microbes.
In conversation with Dr Kay Achenbach, Head of the Data Ethics Professional course at the ODI, Rowenna covers where data harms really come from (mostly not from bad intentions), how cognitive bias shapes assessment, and how the toolkit sits alongside the ODI’s Data Ethics Canvas and consequence scanning. The Q&A takes on live audience questions about sharing sensitive data with public LLMs, handling pushback from senior stakeholders, and whether anyone harmed by data ever gets meaningfully compensated.