Humanoid Training · Part of The Humanoid Group

Training Data and Privacy

Robot training data often captures more than the task. Faces, voices, screens and the inside of people's homes end up in camera and microphone streams. This guide covers the legal basis, consent and handling rules that apply when you record.

By Arjun Rao · Updated October 2026

Why robot recordings count as personal data

Regulation (EU) 2016/679, the General Data Protection Regulation, defines personal data as any information relating to an identified or identifiable natural person, and video of a face or a recording of a voice usually meets that test. A humanoid collecting demonstrations sees far more than the object in its gripper: colleagues walking past, name badges, whiteboards, family photos and papers left on a desk. Head and wrist cameras, the operator's headset feed and any microphones can all hold this material. Treat every stream as personal data unless you have checked otherwise, and work out whose data each recording contains, since that decides whose rights apply and who must be told.

Choosing a lawful basis and handling consent

The GDPR offers several lawful bases for processing, and consent is only one of them. Recital 43 says consent should not be relied on where there is a clear imbalance between the person and the organisation, a situation that often arises between employer and employee, so many organisations look instead to legitimate interests supported by a documented balancing test. In private homes, the householder's consent may be workable, but visitors and children caught on camera have agreed to nothing. Whatever basis you choose, write down your reasoning before collection starts. If footage could reveal health, religion or other special categories of data, stricter conditions apply, and identifying people from faces or voices brings biometric rules into play.

Recording in workplaces

The UK Information Commissioner's Office guidance on monitoring workers gives practical rules that carry over well to robot data collection. It expects employers to tell workers about the extent and nature of monitoring and the reasons for it, and it lists video monitoring among the kinds that call for a data protection impact assessment before work begins. It also advises switching off any audio recording capability by default and enabling it only in exceptional circumstances. For a training programme, that means briefing staff who work in the robot's area, marking where recording happens, offering a route around those zones where possible, and leaving microphones off unless a task truly depends on sound.

Handling and keeping training data

Once footage is collected, the GDPR principles of data minimisation and storage limitation still apply. Blur or crop faces and screens the model does not need, strip audio from episodes where sound plays no part, and keep operator identities in a separate table from the episode files. Limit who can view raw recordings, log access, and encrypt storage and transfers. Set a shorter retention period for raw footage than for processed, de-identified training sets, and delete on schedule rather than when someone remembers. If a contractor or the robot maker receives the data, put the terms in writing, covering permitted uses, storage location and how a person's request to see or erase their data will be passed on and answered.

Sources and further reading

Regulation (EU) 2016/679 (General Data Protection Regulation) — EUR-Lex, Publications Office of the European Union.
The GDPR text: what counts as personal data, the lawful bases for using it, consent rules and the data handling principles.
Employment practices and data protection: monitoring workers (specific considerations for different methods of monitoring) — UK Information Commissioner's Office (ICO).
Regulator guidance on video, audio and biometric monitoring of staff, including audio-off defaults and impact assessments.

Common questions

Can a robot maker use footage recorded in my home?

Only on terms you have agreed. Read the privacy notice and data settings before the robot starts recording, check whether footage ever leaves the device, and look for an option that stops your recordings being used for training. In the EU and UK you can also ask what data is held about you.

Does blurring faces make training video anonymous?

Not necessarily. Body shape, voice, clothing, room layout and file metadata can still identify someone, especially to people who know them. Blurring lowers the risk and is good practice, but keep treating the footage as personal data unless a proper assessment shows nobody can reasonably be identified.

Learning From Demonstration — How humanoids learn skills from human demonstrations and teleoperation data.
Simulation & Sim-to-Real — Why humanoids train in simulation first, and what it takes to cross the gap to the real world.
Evaluation & Readiness — How to judge whether a humanoid is safe, reliable and ready for real work.
Preparing Your Team — The roles, skills and habits your organisation needs before humanoids arrive.
Humanoid Training Services — When to outsource robot data collection, teleoperation and evaluation, and what to ask a provider.
How to Train a Humanoid — A plain-language walkthrough of teaching a humanoid a new task, from demonstration to testing.
Teleoperation for Data — VR, leader-arm and exoskeleton rigs for recording humanoid demonstrations, and what makes a run worth keeping.
Learning to Walk — How reinforcement learning teaches humanoids to walk and balance in simulation before they step out for real.
Training Data and Privacy — Recording people at work or at home to train robots: consent, data protection law and careful data handling.
Choosing First Tasks — How to pick a humanoid's first skills, set pass marks and budget for the demonstrations each one needs.
Safety While Training — Zones, stops and supervision that keep people safe while a humanoid is being taught, tested and debugged.
Latest updates — Teaching robots now takes data operations, simulation and careful testing. These figures and dated updates show where training data comes from, how fast robots learn new tasks and how results are checked.
About us — What we cover, how we source every figure, and who publishes this site.

Visit The Humanoid Group

People also search for robot training data privacy, consent for robot recordings, GDPR robot training data, recording employees for AI training, workplace monitoring robots and data protection impact assessment.