Research ethics

Earth observation is dual-use by nature, and observing the surface of the planet at increasing resolution has privacy consequences that the law has not fully caught up with. Terravek runs an ethics review chaired by the Chief Scientific Officer that every research project passes before work starts.

What the review covers

1. Dual-use potential. Could the capability, as built, be applied to targeting or surveillance of people? If so, can it be constrained by design rather than by policy?
2. Privacy. Does the work involve resolution, revisit rate or inference that could identify individuals or their activity? Terravek does not work at resolutions where individual people are resolvable, and does not build models whose purpose is to infer individual behaviour from land use.
3. Consent and community impact, where fieldwork involves smallholders or communities. The AGRISENSE East Africa programme has an enrolment process, not an assumption.
4. Publication risk. Whether publishing the method creates a capability whose harms outweigh its benefits. This has deferred publication twice; it has never blocked one permanently.
5. Environmental impact of the research itself, principally compute.

Who sits on it

The review is chaired by the Chief Scientific Officer, Dr. Élodie Marchand, with the General Counsel, the Data Protection Officer, the Head of Terravek Labs and one rotating member from outside Labs. It meets monthly and its decisions are recorded.

What it has decided

We publish a summary of decisions annually in the sustainability report rather than the decisions themselves, because several involve customer or partner confidentiality. The categories in 2025 were: 11 approved unconditionally, 4 approved with conditions, 2 deferred pending redesign, 1 declined.

Relationship to the commercial dual-use review

This is a research ethics review, not the commercial screening applied to customer acquisition requests. That is a separate process run by the General Counsel — dual-use and export control policy. The two share a chair deliberately, so that a capability declined in research does not arrive later as a customer request.