Biometric and Emotion-Inference Addendum
SHA-256 6ff4166a927e4ced54f8cc9b6d51eebdcf08e9edb8e2f658ab1eb1b5e3414464
In this document
This Biometric and Emotion-Inference Addendum ("Addendum") forms part of the Agreement between Neurologyca Technologies Ltd. ("Neurologyca", "we", "us") and the Customer. It sets out what the Customer must do before and while its applications capture End Users with a camera or a microphone through the Services, and what Neurologyca does with that data. Capitalized terms not defined here have the meaning given in the Neurologyca DevLab Terms of Service. This Addendum supplements the Data Processing Addendum and the Acceptable Use Policy; where they conflict on biometric data or emotion inference, the provision that better protects End Users prevails.
1. Scope
1.1 This Addendum applies to every capture of an End User with a camera or a microphone made through the Services, whether live or deferred, including captures in the Customer's applications and in its tests. In the Playground, section 3.1 and the DevLab Privacy Notice apply.
1.2 It applies in particular to transcription of speech, to pulse estimation from the face (rPPG), to the optional audio channel, and to deferred processing, in which video with voice is recorded and processed later.
1.3 It applies whether or not a given law classifies the data as biometric. The Customer treats all data covered by section 2 as if it were biometric data.
2. Definitions
"Biometric Data" means any image, video or voice recording of an End User, and any Signal derived from the End User's face, voice or body, including facial landmarks, expression coefficients, head pose, gaze, voice features and pulse signals, and any data that a Biometric Law classifies as biometric identifiers, biometric information or similar.
"Emotion Inference" means any Indicator or Interpretation that estimates an End User's emotional, affective, cognitive or physiological state, including stress.
"Biometric Laws" means every law that regulates the collection, use or retention of biometric data or health data, including the Illinois Biometric Information Privacy Act (BIPA), the Texas Capture or Use of Biometric Identifier Act, Washington's biometric identifier law (RCW 19.375) and My Health My Data Act, the California Consumer Privacy Act, the GDPR and the UK GDPR, the Swiss Federal Act on Data Protection, and Mexico's Federal Law on the Protection of Personal Data Held by Private Parties.
3. Roles
3.1 The Customer decides whether, why and how its applications capture End Users. For Biometric Data processed through Engine, the Hosted Processor and the Interpretation, the Customer is the controller or the business, and Neurologyca acts as its processor or service provider under the Data Processing Addendum. In the Playground, where Authorized Users capture themselves or consenting colleagues to evaluate the Services, Neurologyca is the controller, as described in the DevLab Privacy Notice.
3.2 The Customer's notices and consents may identify Neurologyca as the Customer's service provider. They are given and obtained by the Customer, on its own behalf.
3.3 The Services are not designed to identify or verify people, and the Customer must not use them to do so.
4. How the Services process Biometric Data
4.1 Live capture
In live capture, the SDKs process the camera image on the End User's device and discard each frame. No video or audio leaves the device. The Signals computed on the device, such as facial landmarks, expression coefficients and head pose, are sent through the Customer's backend to Engine, which produces the Indicators. Optional channels, such as gaze, voice features and pulse, are sent only when the Customer enables them. The Documentation lists exactly what each channel sends.
4.2 Transcription
When the Customer enables transcription, the SDKs open the microphone. In live capture, speech is transcribed on the device and only the text is sent. In deferred processing, the audio travels inside the recording and is transcribed on the server. The transcript is the literal text of what was said while a step was open, including anything said by other people nearby.
4.3 Pulse
When the Customer enables the pulse channel, a cardiac signal is estimated from the color of the face and sent as Signals. Pulse estimates are not medical measurements.
4.4 Deferred processing
In deferred processing, a Preview Feature, the End User's device records video with audio, keeps it locally until it is uploaded, and uploads it to storage chosen by the Customer, from which the Hosted Processor or the Customer's own processor replays it to produce the Indicators. Where a recording is held in DevLab, Neurologyca deletes it when the processing job reaches the interpreting or succeeded state, when the job finally fails, or within 24 hours at most. Retention by Engine and by the Hosted Processor is as stated in the Documentation. Deleting recordings in the Customer's own storage is the Customer's responsibility.
4.5 Interpretation
The Interpretation is generated by a third-party large language model listed in Subprocessors, which receives the Indicators and, when transcription is enabled, the transcript. At the effective date of this Addendum that provider is Google, through the paid tier of the Gemini API, whose terms state that Google does not use prompts or responses to improve its products (Gemini API Additional Terms of Service, as of November 3, 2026, https://ai.google.dev/gemini-api/terms). The Customer should use pseudonymous identifiers, never names, for End Users.
4.6 No training
Neurologyca does not use Biometric Data or Emotion Inferences to train models unless the Customer agrees to it in a Written Agreement.
5. Notice before capture
5.1 Before any capture, the Customer gives each End User a clear notice, in a place they will see, that describes at least: that their image and, where applicable, voice will be captured and analyzed with artificial intelligence; what Biometric Data is processed, what stays on the device and what is sent, to whom and where; the purposes; whether a third-party language model generates an Interpretation; how long each kind of data is kept; their rights; and how to refuse or stop at any time.
5.2 The notice must also meet every specific requirement of the applicable law, including the information that an emotion recognition system is in operation where Article 50(3) of the EU AI Act or a similar law applies.
5.3 The model text in section 14 may be adapted for this purpose.
6. Consent
6.1 The Customer obtains every consent that the applicable law requires before the capture starts, and never after it.
6.2 Consent must be in writing where the law requires it. This includes a written release under BIPA before collecting any biometric identifier or biometric information of a person in Illinois, consent under the Texas and Washington biometric laws, explicit consent under Article 9 of the GDPR and the UK GDPR where special categories of data are processed, and express written consent for sensitive personal data under Mexican law. An electronic signature or an affirmative electronic act is valid where the law accepts it.
6.3 Consent must be specific, informed and freely given, must be separate from acceptance of the Customer's general terms where the law requires it, and must cover transcription and deferred recording separately when the Customer uses them.
6.4 The End User must be able to withdraw consent and stop the capture at any time, as easily as they gave it. The Customer stops processing and deletes the data concerned as the law requires.
6.5 The Customer keeps evidence of each notice and consent for as long as the law requires and provides it to Neurologyca on reasonable request.
7. Retention and destruction
7.1 The Customer publishes a written retention and destruction policy for Biometric Data that is available to the public where the law requires it.
7.2 Where BIPA applies, that policy requires permanent destruction of biometric identifiers and biometric information when the initial purpose of collection has been satisfied or within three years of the individual's last interaction with the Customer, whichever occurs first. The Customer applies any shorter period required by other applicable law.
7.3 The Customer complies with its policy, instructs Neurologyca to delete data held through the Services using the means described in the Documentation, and deletes any copy in its own systems.
8. No sale or profit
8.1 The Customer must not sell, lease, trade or otherwise profit from Biometric Data or Emotion Inferences, and must not disclose them except where the applicable law permits and, where required, with the End User's consent.
8.2 Neurologyca does not sell, lease, trade or otherwise profit from Biometric Data or Emotion Inferences, and discloses them only to the subprocessors listed in Subprocessors to provide the Services, or where the law requires it.
9. Health data
9.1 Pulse estimates and inferences of stress or other physiological or mental states may be health data under the law of some places, including special-category data under Article 9 of the GDPR and consumer health data under Washington's My Health My Data Act.
9.2 The Customer treats them as health data unless it has determined, for its specific use and jurisdictions, that they are not. Where they are health data, the Customer obtains explicit consent, or relies on another legal basis that the law allows, before processing them, and meets every additional requirement of the applicable law.
9.3 Pulse estimates and Emotion Inferences are not medical devices or medical results and must not be used as such, as the Acceptable Use Policy provides.
10. Security
10.1 The Customer stores, transmits and protects Biometric Data with a reasonable standard of care in its industry, and in a manner at least as protective as the way it protects its other confidential and sensitive information.
10.2 The Customer keeps its API Keys on its backend, never in a browser or a mobile application, and controls who in its organization can access Biometric Data.
10.3 Neurologyca applies the security measures described in the Data Processing Addendum.
11. Impact assessments and assistance
11.1 The Customer carries out a data protection impact assessment, or any equivalent assessment, wherever the law requires it, before the capture starts.
11.2 Neurologyca provides reasonable assistance, including information about how the Services process Biometric Data, as described in the Data Processing Addendum.
12. End-user requests
The Customer handles End Users' requests to access, correct or delete their data or to withdraw consent. Neurologyca assists the Customer with requests concerning data held through the Services as described in the Data Processing Addendum, and forwards to the Customer any request it receives directly from an End User of the Customer.
13. Verification, non-compliance and indemnity
13.1 On reasonable request, the Customer gives Neurologyca information that shows it complies with this Addendum, including evidence of its notices, consents and retention policy.
13.2 If the Customer does not comply with this Addendum, it must stop capturing End Users until it does, and Neurologyca may restrict or suspend the affected features as the Acceptable Use Policy and the Terms provide.
13.3 The Customer defends, indemnifies and holds harmless Neurologyca, its affiliates and their personnel against any third-party claim, fine or penalty arising from the Customer's failure to comply with this Addendum or with Biometric Laws, including reasonable legal fees, as set out in the Terms.
14. Model notice and consent text
14.1 How to use the template
The following text is a template that the Customer may adapt to its application. It is not legal advice and does not, by itself, guarantee compliance with any law. Before using it, the Customer states its specific purpose at the start of the notice, removes the paragraphs that do not apply (they are marked "if you use…"), adds anything its jurisdictions require, replaces the placeholder "[Your company name and contact]" with its own legal name and contact details, and has the result reviewed by its own counsel. The consent checkboxes must be unticked by default and must be shown before the camera or microphone is turned on.
14.2 Template
How we analyze your camera and voice
[Your company name and contact] ("we") would like to analyze your face and, if you agree, your voice during this session, using artificial intelligence provided by our service provider, Neurologyca. This is an emotion recognition system: it estimates indicators such as attention, emotional expression and stress. These are probabilistic estimates. They are not a diagnosis, they do not detect lies and they are not used to identify you.
What is captured and where it goes. Your camera image is processed on your own device and is not sent anywhere. Your device computes measurements of your face, such as the position of facial points and expressions, and sends only those measurements to our service provider, which turns them into indicators.
(If you use transcription.) Your microphone is used to transcribe what you say during the session. The text of what you say, including anything said by people near you, is sent with the measurements.
(If you use the pulse channel.) A pulse signal is estimated from the color of your face. It may be considered health data.
(If you use deferred processing.) Your device records video with sound of the session and uploads it to our servers so that it can be analyzed afterwards. The recording is deleted once it has been analyzed.
Written interpretation. The indicators and, if your voice is transcribed, the text of what you said are sent to an artificial-intelligence language model operated by a third-party provider, which writes an interpretation of the session. That provider does not use them to train or improve its products.
Why we use it. We use the results only for the purpose described above, and never to make a decision about you without a person reviewing it.
How long we keep it. We keep your data only as long as needed for that purpose, as described in our retention policy, and destroy biometric data when that purpose is met and, in any case, within three years of your last interaction with us, or sooner where the law requires.
We do not sell it. We do not sell, lease or trade your biometric data or the indicators about you.
Your rights. You can ask us to access, correct or delete your data and withdraw your consent at any time, as easily as you gave it, by contacting [Your company name and contact]. Depending on where you live, you may also have the right to complain to a data protection authority.
You can say no. Taking part is voluntary. You can refuse now, or stop the session at any time, without giving a reason.
- Checkbox: I have read this notice and I consent to the analysis of my face described above, including the capture and use of my biometric data.
- Checkbox (if you use transcription): I consent to the capture and transcription of my voice.
- Checkbox (if you use the pulse channel or infer stress): I consent to the processing of data that may be health data, as described above.
- Checkbox (if you use deferred processing): I consent to the recording and upload of video with sound of this session.
15. Precedence
This Addendum prevails over the Terms and the Preview Terms on matters of Biometric Data and Emotion Inference. It does not reduce the Customer's obligations under the Acceptable Use Policy or under the law, nor Neurologyca's obligations under the Data Processing Addendum.
Versions
| Version | In effect | What changes | SHA-256 |
|---|---|---|---|
| 1.1 current | Nov 17, 2026 | What changesA new section 4.7 sets out what is Biometric Data in each mode: in live capture, the camera image and, where enabled, the audio, which do not leave the device, and the facial landmarks and other Signals computed on it; in deferred processing, also the uploaded video with the face and voice. A new section 6.6 requires prior written consent before a deferred recording starts wherever the law requires it. The model notice in section 14 now says that the results are an estimate and not professional help, and that the interpretation is generated by artificial intelligence. | 0b4c343d0d6c49128438d99a62ac2c505bfaef7516467627d558dd2b69617441 |
| 1.0 | Nov 3, 2026 | 6ff4166a927e4ced54f8cc9b6d51eebdcf08e9edb8e2f658ab1eb1b5e3414464 |