Legal Overview
SerenaChat is a symbolic affective computing system developed by Beyond The Box. This page consolidates all legal information, including Terms of Service, API terms, data handling, acceptable use, privacy principles, and your rights as a user.
1. Terms of Service
1.1 Acceptance of Terms
By accessing or using SerenaChat, its API, demo interface, documentation, website, or any related services provided by Beyond The Box, you acknowledge that you have read, understood, and agree to be bound by these Terms of Service. If you do not agree with these terms, you must discontinue use of the platform immediately.
If you are using SerenaChat on behalf of an organisation, you confirm that you are authorised to agree to these terms on its behalf.
1.2 Description of Service
SerenaChat is a symbolic emotional analysis engine that converts input text into a structured emotional vector. The platform provides:
- Deterministic emotional vector computation
- Emotion classification (primary, secondary, tertiary)
- A public demo endpoint for testing
- API access for developers and businesses
- Supporting dashboards, documentation, and integrations
SerenaChat is entirely symbolic. It does not operate as a neural network, does not store user-submitted text, and does not use any form of user content for training.
1.3 Modifications and Availability
Beyond The Box may modify, add, or remove features of SerenaChat at any time. This includes, but is not limited to:
- API features or endpoints
- Rate limits or pricing tiers
- Dashboard functionality
- Backend behaviour and internal processes
- Demo system limitations
Where possible, we will provide reasonable notice for material changes. Minor changes or improvements may occur without notice. SerenaChat may also be temporarily unavailable due to maintenance, updates, or technical issues.
1.4 Eligibility
You must be at least 13 years old to use SerenaChat. If you are under 18, you must use the service only under supervision from a parent, guardian, or authorised adult.
1.5 Account Registration
Certain features (such as API keys or usage dashboards) may require an account. You are responsible for maintaining the confidentiality of your credentials and for any activity occurring under your account. Beyond The Box is not liable for unauthorised access resulting from failure to secure your login details.
1.6 Service Restrictions
You may not:
- Use automated scripts to abuse the demo endpoint
- Attempt to reverse engineer, extract, or replicate the emotional mapping system
- Interfere with system integrity or security
- Resell or sublicense the service without permission
- Use SerenaChat outputs as the sole basis for high-risk decisions
1.7 Third-Party Integrations
You may integrate SerenaChat into your own products or tools, provided such integrations comply with these Terms and do not misrepresent or expose the internal workings of the emotional engine. Beyond The Box is not responsible for issues arising from third-party integrations or systems outside our control.
1.8 Suspension and Termination
Beyond The Box reserves the right to suspend or terminate access to SerenaChat if:
- You violate these Terms
- You misuse or attack the service
- Your usage poses security or stability risks
- Fraudulent or abusive behaviour is detected
Termination may occur without prior notice if necessary to protect the service or users.
1.9 No Warranties
SerenaChat is provided “as is” and “as available”. Beyond The Box makes no warranties, whether express or implied, regarding:
- Accuracy of emotional classification
- Uninterrupted availability
- Fitness for any particular purpose
- Error-free operation
1.10 Governing Law
These Terms are governed by the laws of the United Kingdom, without regard to conflict of law principles. Any disputes shall be resolved through binding arbitration unless a separate enterprise contract specifies alternative terms.
1.11 Updates to These Terms
We may update these Terms from time to time. The “Last Updated” date at the bottom of this page will reflect the most recent revision. Continued use of the service constitutes acceptance of the updated Terms.
2. Acceptable Use Policy
This Acceptable Use Policy ("AUP") outlines activities that are prohibited or restricted when using SerenaChat, its API, demo interface, documentation, dashboards, or any related services operated by Beyond The Box.
2.1 Prohibited Activities
You agree not to use SerenaChat for:
- Illegal activities, including fraud, harassment, or distribution of harmful content.
- Autonomous weapons, targeting systems, or military-grade surveillance tools.
- Detection, manipulation, or profiling of protected characteristics (race, religion, health status, etc.).
- High-risk decision making without appropriate human oversight (medical triage, legal decisions, etc.).
- Building competitive replicas of SerenaChat via reverse engineering or extraction of system logic.
- Exceeding rate limits or attempting to bypass usage caps.
- Scraping, automated abuse, or load testing of the demo endpoint.
- Any activity intended to degrade, disrupt, or impair the service.
2.2 Demo Environment Restrictions
The demo endpoint is strictly for human testing and evaluation. Automated or bulk use, production traffic, or integration into commercial systems using the demo endpoint is prohibited.
2.3 Security Obligations
You must not interfere with the security, integrity, or availability of SerenaChat. This includes prohibitions on:
- Attempting unauthorised access to systems, accounts, or data
- Sending malware, exploits, or harmful payloads
- Probing, scanning, or stress-testing infrastructure without permission
2.4 Enforcement
Beyond The Box reserves the right to monitor usage for compliance and may suspend or terminate access without notice if violations are detected.
3. API Terms
3.1 API Key Usage
Access to the SerenaChat API requires an authenticated key linked to your account. You must keep API keys confidential and may not share, resell, or misuse them.
3.2 Rate Limits
Each plan includes specified monthly and per-minute request limits. Exceeding these limits may trigger throttling, temporary blocking, or additional charges for enterprise plans where applicable.
3.3 Allowed Integrations
You may integrate the API into your products, services, internal tools, or research systems, provided such integrations:
- Do not misrepresent SerenaChat output as clinical, diagnostic, or legally binding
- Do not create derivative models attempting to replicate Serena's logic
- Respect all applicable data and privacy laws
3.4 Logging and Data Retention
SerenaChat does not store user-submitted text. The system stores only lightweight, non-content metadata (timestamps, token counts, and error codes) for short-term operational debugging, unless an enterprise contract specifies otherwise.
3.5 SLA and Availability
Public users receive best-effort uptime. Enterprise plans may include:
- Guaranteed uptime SLAs
- Support or incident response windows
- Dedicated or private endpoints
3.6 Misuse of the API
Activities such as load testing, security probing, or intentional destabilisation of API infrastructure are prohibited without explicit permission.
4. Privacy Principles
SerenaChat is designed from the ground up with privacy and minimisation in mind. The engine does not store, log, or train on user input text.
4.1 No Storage of Input Text
Text sent to the API is processed **in memory only** and is immediately discarded once the emotional vector is produced.
4.2 No Training on User Data
SerenaChat is symbolic and therefore does not learn from or incorporate user text into any model.
4.3 Metadata Handling
Minimal metadata may be retained temporarily for debugging:
- Request timestamps
- API key identifier
- Latency and error diagnostics
No personal data is extracted from input text.
4.4 GDPR & International Compliance
SerenaChat follows GDPR-compliant principles:
- Data minimisation
- Purpose limitation
- Transparency
- User control in enterprise contexts
4.5 Enterprise Privacy Mode
Enterprise deployments may:
- Disable all logging entirely
- Run Serena locally within a private VPC
- Route data exclusively through customer infrastructure
5. Intellectual Property
SerenaChat, including its symbolic emotional engine, classification system, vector mappings, source code, trademarks, design, documentation, and branding elements, is the exclusive property of Beyond The Box.
5.1 No Replication
You may not copy, reverse engineer, extract, or attempt to recreate the underlying symbolic logic, emotional rules, or codebase.
5.2 No Redistribution
You may not redistribute the SerenaChat system, its outputs, or any portion of the emotional engine without explicit written permission.
5.3 Use of Brand Assets
“SerenaChat,” “Serena,” and all associated branding may only be used under permitted guidelines or with written approval.
6. Limitation of Liability
To the maximum extent permitted by law, SerenaChat is provided “as is” and “as available,” without warranties of any kind, express or implied.
6.1 No Guarantee of Accuracy
Emotional classifications, vectors, and system outputs are deterministic but may not always reflect real human emotions or intended meanings.
6.2 No Liability for Damages
Beyond The Box is not liable for:
- Loss of profits or business interruption
- Loss of data or corruption of information
- Misuse or misinterpretation of outputs
- Indirect or consequential damages
6.3 No Professional Use Claims
SerenaChat outputs must not be used as a substitute for mental health, medical, legal, or safety-critical judgement.
7. Data Handling
SerenaChat operates with strict data minimisation principles.
7.1 In-Memory Processing
Input text is not stored. It is processed temporarily in memory, converted into an emotional vector, then immediately discarded.
7.2 Logging Configuration
Public API logs include only non-sensitive metadata. Enterprise customers can customise or disable all logging.
7.3 No Dataset Creation
SerenaChat does not build datasets from user interactions. No text is retained for training, fine-tuning, or analytics.
7.4 Local and On-Prem Deployments
Enterprise users may deploy SerenaChat in isolated environments with:
- No external network dependencies
- Complete control over logs and retention
- Total data sovereignty for compliance needs
9. Responsible AI Policy
SerenaChat is designed to provide emotionally-aware computation through a fully symbolic, deterministic engine. As part of Beyond The Box’s commitment to safe and responsible AI development, we maintain strict guidelines that govern how SerenaChat is built, deployed, and used.
9.1 Ethical Foundations
SerenaChat is developed according to the principles of transparency, safety, and explainability. The system does not generate free-form text or predictions, does not infer protected attributes, and does not take actions autonomously. All outputs are deterministic and fully inspectable.
9.2 Transparency and Explainability
Unlike black-box neural networks, SerenaChat provides:
- Deterministic outputs with no hidden parameters
- A direct mapping from input text to emotional vector
- Full interpretability of the reasoning process
- No stochastic behaviour or hallucination risk
These properties make SerenaChat suitable for regulated industries and applications requiring predictable behaviour.
9.3 Safety and Harm Prevention
SerenaChat is not intended to:
- Replace human judgment in high-risk environments
- Diagnose, treat, or manage mental health conditions
- Make legal, medical, financial, or safety-critical decisions
- Profile users based on sensitive characteristics
Users integrating SerenaChat into products must provide appropriate safeguards, oversight, and human review for any sensitive or consequential applications.
9.4 Privacy Preservation
SerenaChat follows strict data minimisation practices:
- No storage of input text
- No training or optimisation on user data
- No creation of behavioural profiles or user models
- No sharing or selling of personal data
The system processes text exclusively in memory. Enterprise deployments allow customers to enforce their own privacy policies through private or on-premise installations.
9.5 Human Oversight
SerenaChat is designed as a supportive emotional computation tool and must be used with human supervision in workflows involving:
- User wellbeing or mental health
- Customer service escalation decision-making
- Workflow automation or agent behaviour
- Moderation or review pipelines
9.6 Restriction on Sensitive Use Cases
SerenaChat may not be used in:
- Autonomous systems capable of causing harm
- Emotion manipulation at scale (e.g., political persuasion engines)
- Law enforcement or real-time surveillance systems
- Military, defence, or weapons targeting applications
- Detection of mental illness, trauma, or physiological states
9.7 Continual Review
Beyond The Box reviews the ethical and societal impact of SerenaChat on an ongoing basis. As the system evolves (e.g., future language or response-generation modules), we will update this policy to reflect new capabilities, risks, and mitigations.
9.8 Developer Responsibility
Developers integrating SerenaChat into their systems must:
- Comply with all applicable laws and regulations
- Provide transparent disclosures to end users when appropriate
- Avoid deploying SerenaChat in prohibited contexts
- Maintain appropriate safeguards around user privacy
9.9 Reporting Concerns
If you believe SerenaChat is being used in a harmful, unethical, or inappropriate manner,
please report it to:
responsibleai@serenachat.com
This policy will continue to evolve as SerenaChat’s capabilities expand.
10. AI Safety Commitments
SerenaChat is built as a transparent, deterministic, symbolic system designed with safety as a foundational principle. Beyond The Box commits to developing and deploying Serena in ways that reduce harm, prevent misuse, and ensure predictable behaviour.
10.1 No Autonomous Actions
SerenaChat does not make autonomous decisions, initiate actions, or execute commands. Outputs are strictly informational and require human oversight.
10.2 No Neural-Style Hallucinations
SerenaChat produces deterministic emotional vectors using defined symbolic rules. It does not generate or infer facts, opinions, or free-form text, preventing hallucination risk.
10.3 No Behavioural Manipulation
SerenaChat must not be used to manipulate user emotions, influence political views, or shape behaviour at scale without transparent and ethical frameworks.
10.4 Human-in-the-Loop by Default
Any workflow incorporating SerenaChat must maintain human supervision, especially in:
- Mental health or wellbeing contexts
- Customer conflict resolution
- Safety-critical decisions
- Automated agent behaviour
10.5 Continuous Monitoring
Beyond The Box actively reviews potential misuse scenarios and updates safeguards across policy, API design, and access controls.
11. Serena Model Card
11.1 System Overview
SerenaChat is a deterministic, symbolic emotional analysis engine. It maps input text to a 7-dimensional emotional vector, then classifies the closest matching emotional states from an ontology of ~200 emotions.
11.2 Intended Use
Serena is intended for:
- Agent emotional awareness
- Customer support triage
- NPC emotional modelling
- Wellbeing trend tracking (non-diagnostic)
- Analytics, dashboards, and research tools
11.3 Out-of-Scope Uses
Serena must not be used for:
- Mental health diagnosis or treatment
- Legal, financial, or medical decision-making
- Surveillance or identity inference
- Emotion manipulation or political persuasion
- Predicting protected characteristics
11.4 Inputs
Input: raw text. Limitations:
- Outputs depend on clarity of user text
- No inference of context not provided in input
- No world knowledge or memory outside the request
11.5 Outputs
Serena returns:
- A 7D emotional vector
- Primary, secondary, tertiary emotions
- Structured JSON for downstream systems
11.6 Limitations
- Deterministic but not always aligned with human interpretation
- No cultural or linguistic contextual modelling
- Cannot detect sarcasm, idioms, or complex metaphor reliably
11.7 Intended Users
Developers, researchers, agent designers, simulation creators, wellbeing tool builders, and system architects requiring predictable emotional modelling.
12. High-Risk and AGI-Adjacent Use Statement
SerenaChat is not an AGI system and does not approximate AGI-like behaviours. It does not demonstrate autonomous agency, multi-step reasoning, or long-term goal formation. It is a symbolic emotional layer designed to augment other systems.
12.1 Prohibited High-Risk Uses
Serena must not be used in:
- Autonomous weapon systems
- Autonomous vehicles or robotics with safety-critical roles
- Real-time surveillance pipelines
- Life-or-death decision systems
- Emergency response automation
12.2 AI Research Transparency
Serena’s symbolic engine promotes safe AGI research by:
- Providing an interpretable emotional layer
- Allowing controlled emotional modulation in agents
- Enabling emotionally coherent simulations without neural black-box behaviour
12.3 No Foundation Model Properties
SerenaChat does not:
- Learn from user data
- Develop new capabilities over time
- Store memory or experiential knowledge
- Generate text, plans, or actions
13. Security Practices
Beyond The Box employs industry-standard security measures to protect the integrity, availability, and confidentiality of SerenaChat.
13.1 Infrastructure Security
Our security stack includes:
- Encrypted transit (HTTPS/TLS 1.2+)
- Firewall-restricted compute environments
- Strict API authentication via key-based access
- Rate limiting to prevent abuse and DDoS behaviour
13.2 Application Security
- Static code analysis and dependency audits
- Automated linting and test pipelines
- Regular review of emotional mapping logic for edge cases
13.3 Data Security
Serena does not store or retain user text. Enterprise customers may implement their own retention rules.
13.4 Incident Response
In the event of a security incident, Beyond The Box will notify affected enterprise customers promptly and provide remediation guidance.
8. Contact
For legal enquiries, reach us at:
legal@serenachat.com
For enterprise or compliance-specific questions:
enterprise@serenachat.com
Last Updated: January 2025