This Emotional API acceptable use policy (“API Policy”) sets out how you may and may not use the Emotional API, including its endpoints, outputs, scripts, test tokens and documentation (together, the “Emotional API”).
The Emotional API is provided by Sentient Machines Limited (“we” or “us”), registered in England and Wales under company number 10470936, with our registered office at 5th Floor, 16 Great Queen Street, Covent Garden, London, WC2B 5DG.
Who this policy applies to
This API Policy applies to:
Customers. Every organisation that requests a test token, holds an API token or otherwise accesses the Emotional API (“Customer” or “you”), and everyone acting on a Customer’s behalf, including employees, contractors and integration partners.
End users. Every person or organisation that uses the Emotional API, or any Output, through a Customer’s own products or services (“End Users”).
In this API Policy, “Individuals” means the people whose calls, voices, chats or emails are analysed using the Emotional API, such as callers, customers and agents.
You are responsible for making sure that everyone who uses the Emotional API or its Outputs through you complies with this API Policy. If you make Emotional API functionality or Outputs available to End Users, you must bind them to terms at least as protective as this API Policy.
This API Policy applies in addition to our Acceptable Use Policy (which covers use of our website), our Terms of Use, our Data Privacy Policy and any agreement you have signed with us. If this API Policy conflicts with a signed agreement, the signed agreement takes precedence, except that the prohibited uses in this API Policy apply in all cases.
What the outputs are
The Emotional API produces automated, probabilistic estimates from recordings and text, such as emotion proportions (frustration, anxiety and calm), the Emotional Risk Index, sentiment, vulnerability indicators, behaviour and performance scores, and voice-authenticity metrics (“Outputs”).
Outputs describe signals detected in a recording or text. They are not facts about a person’s inner state, character, honesty or health, and they can be wrong. You must treat them as one input among others and must not rely on them alone for any decision that affects an Individual.
Permitted uses
You may use the Emotional API and its Outputs to understand and improve conversations and to help people, for example to:
improve customer experience and service quality;
coach and train agents and teams, subject to the jurisdiction rules below;
identify callers who may need extra support or a different approach, and route them to appropriately trained people;
prioritise complaints, follow-ups and quality reviews;
flag possible fraud or impersonation so that further verification can take place; and
produce aggregated insight and reporting about conversations.
Prohibited uses
You must not use the Emotional API or any Output, or allow anyone else to use them:
To manipulate or pressure people. To exploit an Individual’s emotional state, distress or vulnerability in order to pressure them into a purchase, contract, payment, consent or any other decision, including timing, scripting or targeting persuasion at moments of detected emotional difficulty.
For credit, insurance, pricing or employment decisions. As an input to any decision about an Individual’s eligibility for, or the terms or price of, credit, lending, insurance or any other product or service; or any employment decision, including recruitment, pay, promotion, disciplinary action or dismissal; or any other decision that produces legal or similarly significant effects for an Individual. Agent and team Outputs may be used for coaching, training and quality improvement only.
As lie detection or diagnosis. To present, market or rely on Outputs as detecting lies, deception or truthfulness, or as diagnosing or assessing any physical or mental health condition.
To infer sensitive characteristics. To infer or attempt to infer an Individual’s racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, sexual orientation or health.
Unlawfully or covertly. In breach of any applicable law, including data protection, privacy, call-recording, consumer-protection, equality and employment law, or to monitor Individuals without the notice and lawful basis that law requires.
To discriminate. To treat any Individual or group unfairly or unlawfully on the basis of Outputs.
To misuse the service. To share tokens outside your organisation, circumvent rate limits, scopes or permissions, reverse engineer the Emotional API, or use Outputs to build or train a competing emotion-analysis or voice-authenticity service.
Vulnerability indicators
Outputs that suggest an Individual may be vulnerable, such as vulnerability tags, vulnerability labels and vulnerability flags in customer profiles (“Vulnerability Indicators”), require particular care.
You may use Vulnerability Indicators only to support the Individual concerned, for example to offer additional help, adjust how you communicate with them, involve a specialist team, or meet your regulatory obligations to treat vulnerable customers fairly.
You must not use Vulnerability Indicators for pricing, marketing or sales targeting, or for ranking or prioritising Individuals by commercial value, or for any of the prohibited uses above.
Voice authenticity and deepfake outputs
Voice authenticity Outputs, including deepfake metrics, deepfake risk scores and risk bands, are risk indicators. They are designed to trigger further verification, not to replace it. You must not use them:
as proof of a person’s identity, or as a standalone method of identifying or authenticating anyone;
as proof that a recording is genuine or fake; or
as the sole basis for refusing a service, permanently blocking a transaction, accusing anyone of fraud or reporting anyone to the authorities, without further verification and human review.
Jurisdiction-specific rules
You are responsible for using the Emotional API in a way that complies with the laws that apply to you and to the Individuals whose recordings you submit. We may restrict features or apply region-specific processing to help you and us comply with those laws.
In particular, the EU Artificial Intelligence Act prohibits the use of AI systems to infer the emotions of people in the workplace or in education institutions, except for medical or safety reasons. Where that prohibition applies, for example where recordings include agents or other workers located in the European Union, you must not use the Emotional API to infer those people’s emotions. You must tell us before submitting such recordings so that we can apply the processing configuration that removes emotion inference for those people, and you must not attempt to bypass that configuration.
Your responsibilities
When you use the Emotional API, you must:
have a lawful basis for recording and analysing calls and messages, and only submit recordings and text that you are entitled to process;
tell Individuals clearly that their calls are recorded and that their calls and messages may be analysed by automated systems, including for emotion and voice authenticity, wherever the law requires it;
make sure a suitably trained person reviews Outputs before any decision that affects an Individual;
keep your API tokens secure and confidential, and tell us immediately if a token may have been exposed; and
tell us promptly if you become aware of any use of the Emotional API that breaches this API Policy.
Monitoring, suspension and termination
We may monitor use of the Emotional API, including request metadata and patterns of use, to check compliance with this API Policy.
If we believe this API Policy has been breached, we may take any action we consider appropriate, including:
suspending or revoking API tokens, scopes or permissions;
requiring you to stop a particular use case or integration;
withdrawing access to the Emotional API, temporarily or permanently; and
any of the actions described in our Acceptable Use Policy.
We exclude liability for actions taken in response to breaches of this API Policy.
Reporting concerns
If you believe the Emotional API is being misused, please email hello@sentientmachines.tech.
Changes to this policy
We may revise this API Policy at any time by amending this page. Where practicable, we will notify Customers of material changes by email before they take effect. You are expected to check this page from time to time, as the version in force applies to your use of the Emotional API.
Sentient Machines Limited · 5th Floor, 16 Great Queen Street, Covent Garden, London, WC2B 5DG