Back to Commitments

Eddy AI Wellbeing Alerts and Escalation Policy (For Parents, Guardians, and Supervising Organisations)

1. Purpose of This Policy

This policy sets out how the Eddy AI wellbeing alerts system operates, including how alerts are generated, how they are communicated, and where responsibility sits for monitoring, interpretation, and action.

The purpose of this policy is to ensure that all parties clearly understand:

  • the role and limitations of Eddy AI
  • how wellbeing alerts function in practice
  • who is responsible for responding to potential concerns
  • the boundaries between AI generated insights and human safeguarding duties

This policy applies to all individuals and organisations who use, oversee, or are responsible for users of Eddy AI. This includes, but is not limited to:

  • parents and guardians
  • schools, colleges, and educational institutions
  • charities and third sector organisations
  • local authorities and public sector bodies
  • employers, training providers, and programme operators
  • any individual or organisation with supervisory, pastoral, or duty of care responsibilities

Eddy AI is designed to provide assistive, technology enabled insight to support engagement, wellbeing awareness, and structured interaction. It is not designed to act as a safeguarding system or decision maker.

Eddy AI must be understood as:

  • a support tool, not a monitoring service
  • a signal generating system, not an assessment or diagnostic tool
  • a complement to human oversight, not a replacement for it

Eddy AI must not be relied upon as:

  • a safeguarding authority or reporting mechanism
  • a clinical or mental health service
  • a crisis detection or emergency response system
  • a substitute for trained professionals, statutory processes, or organisational safeguarding procedures

All users and supervising parties must recognise that wellbeing alerts generated by Eddy AI are:

  • indicative only
  • based on automated and probabilistic processes
  • subject to limitations, inaccuracies, and delays

Accordingly, this policy establishes that:

  • Eddy AI does not assume responsibility for user welfare or safeguarding outcomes
  • all duty of care remains with the supervising organisation or individual
  • all safeguarding decisions must be made by appropriately trained humans

By using Eddy AI, or by enabling access to it, supervising organisations, parents, and guardians acknowledge and accept the purpose, scope, and limitations outlined in this policy.

2. Nature of the System

Eddy AI uses automated and rules based processes to identify potential indicators that a user may be experiencing emotional, behavioural, or wellbeing related difficulty during their interaction with the platform.

These processes are designed to support human oversight by surfacing potential concerns for review. They do not involve continuous human monitoring and do not constitute professional judgment.

2.1 Core System Function

Eddy AI operates as an assistive signal detection system, using a combination of:

  • rule based thresholds
  • pattern recognition
  • AI driven language analysis

The system analyses user interactions within the platform to identify possible indicators of concern, which may then generate alerts for supervising individuals or organisations.

2.2 Sources of Detection

The system may generate indicators or alerts based on one or more of the following inputs:

a) Self Reported Wellbeing Data

  • User submitted scores or assessments relating to emotional, social, or mental wellbeing
  • Threshold based triggers (for example low scoring ranges)

b) Language and Behavioural Pattern Recognition

  • Analysis of written or conversational inputs
  • Identification of patterns aligned to recognised behavioural or therapeutic frameworks
  • Detection of indicators such as distress language, withdrawal, avoidance, or escalation in tone

c) AI Generated Concern Flags

  • Automated classification of user content
  • Identification of phrases, themes, or expressions that may indicate risk, vulnerability, or need for support
  • Flagging of potential safeguarding related disclosures based on predefined criteria

2.3 Nature of Outputs

All outputs generated by Eddy AI, including:

  • alerts
  • scores
  • flags
  • classifications
  • suggested interpretations

are:

  • indicative only
  • probabilistic in nature
  • generated through automated processes

These outputs are designed to support awareness, not to provide conclusions.

2.4 System Limitations

Eddy AI has inherent technical and contextual limitations. As such:

  • The system does not understand full context in the way a trained human professional does
  • Inputs may be misinterpreted, incomplete, or ambiguous
  • The system may generate:
    • false positives (flagging non issues)
    • false negatives (failing to detect genuine issues)
  • Outputs may be:
    • delayed
    • inaccurate
    • incomplete
    • taken out of context

There is no guarantee that all risks, behaviours, or safeguarding concerns will be identified.

2.5 No Clinical or Safeguarding Determination

Eddy AI does not:

  • provide medical or mental health diagnosis
  • perform psychological or clinical assessments
  • determine safeguarding risk or status
  • make decisions about intervention or escalation

No output from the system should be interpreted as:

  • a diagnosis
  • a formal assessment
  • a safeguarding finding
  • a risk classification
  • professional advice

2.6 Role of Human Oversight

All outputs must be:

  • reviewed by a responsible adult or organisation
  • interpreted using professional judgment
  • considered alongside wider contextual information

All decisions relating to:

  • safeguarding
  • escalation
  • intervention
  • support

must be made by appropriately trained individuals within the supervising organisation.

2.7 Acceptance of System Nature

By using Eddy AI, all supervising organisations, parents, guardians, and oversight users acknowledge that:

  • the system is assistive only
  • outputs are inherently limited and non definitive
  • reliance must not be placed solely on the system
  • full responsibility for interpretation and action remains with them

3. Alerts and Notifications

Where enabled and configured by a supervising organisation, Eddy AI may generate alerts based on system detected indicators of potential concern.

Alerts are designed to support awareness and prompt human review. They do not constitute decisions, actions, or safeguarding outcomes.

3.1 Alert Generation

Alerts may be generated when predefined system conditions are met, including:

  • low wellbeing scores
  • detection of behavioural or linguistic patterns
  • AI identified concern indicators within user interactions

All alerts are generated automatically based on system logic and thresholds configured within the platform.

3.2 Alert Categories

Alerts are categorised to support prioritisation by supervising staff:

a) Warning Alerts

  • Indicate lower level or emerging concerns
  • May include mild distress signals or reduced wellbeing scores
  • Require review but not necessarily immediate action

b) Critical Alerts

  • Indicate higher level or potentially urgent concerns
  • May include significant distress indicators or high risk language patterns
  • Require prompt attention by the supervising organisation

Alert categorisation is indicative only and must not be treated as a definitive assessment of risk.

3.3 Alert Delivery

Alerts are communicated to the supervising organisation through configured channels, which may include:

  • Email notifications sent to designated staff or responsible individuals
  • Real time or near real time updates within administrative dashboards
  • Internal platform alerts visible to authorised users

Recipients of alerts are determined by the organisation’s configuration settings.

3.4 Alert Content

Alerts may include:

  • a summary of the trigger or indicator
  • contextual excerpts or references (where appropriate)
  • suggested guidance or prompts for review

All guidance provided is:

  • general in nature
  • non binding
  • intended to support, not direct, decision making

3.5 Organisational Control and Configuration

The supervising organisation is solely responsible for:

  • enabling or disabling alerts
  • determining which triggers are active
  • configuring thresholds and rules (where applicable)
  • selecting who receives alerts
  • ensuring alerts are monitored and acted upon

Ambertrace Labsdoes not control how alerts are configured, received, or responded to once the system is in use.

3.6 No Monitoring or Response Guarantee

Eddy AI does not guarantee that:

  • alerts will be generated in all relevant situations
  • alerts will be delivered without delay
  • alerts will be seen, reviewed, or acted upon by staff

The system does not provide continuous monitoring or active supervision.

3.7 No Direct External Notification

Eddy AI does not:

  • contact emergency services
  • notify parents or guardians directly
  • notify external agencies or authorities

Any such communication must be carried out by the supervising organisation in accordance with its own policies and legal obligations.

3.8 No Safeguarding Action by the System

Eddy AI does not:

  • initiate safeguarding procedures
  • escalate cases to authorities
  • intervene in user situations
  • make decisions regarding welfare or risk

All safeguarding action, escalation, and decision making remains the responsibility of the supervising organisation.

3.9 Requirement for Human Review

All alerts must be:

  • reviewed by appropriately trained staff
  • assessed using professional judgment
  • considered alongside broader contextual information

No alert should be relied upon in isolation.

3.10 Acceptance of Alert Limitations

By using Eddy AI, supervising organisations and oversight users acknowledge that:

  • alerts are automated and probabilistic
  • alerts may be inaccurate, incomplete, or delayed
  • alerts are assistive signals only
  • full responsibility for interpretation and response remains with them

4. No Monitoring or Response Guarantee

Eddy AI does not provide continuous, real time, or human monitoring of user activity, behaviour, or wellbeing.

The platform operates as an automated, assistive system and must not be relied upon as a supervision, safeguarding, or monitoring service.

4.1 Absence of Continuous Monitoring

Eddy AI:

  • does not involve live human oversight of user interactions
  • does not monitor users on an ongoing or uninterrupted basis
  • does not review all content in real time Any detection of potential concern is dependent on automated processes and configured system conditions.

4.2 No Guarantee of Alert Generation

Supervising organisations and users must assume that:

  • alerts may not be generated in all relevant or high risk situations
  • some behaviours or disclosures may not meet system thresholds
  • certain risks may not be identifiable by automated processes

Eddy AI does not guarantee that all concerns, risks, or safeguarding issues will be detected.

4.3 Limitations in Accuracy and Completeness

Alerts and outputs generated by Eddy AI may be:

  • inaccurate
  • incomplete
  • delayed
  • misinterpreted

The system may:

  • fail to identify genuine concerns
  • generate alerts where no real issue exists

All outputs must therefore be treated with caution and reviewed by a responsible human.

4.4 No Guarantee of Timely Delivery or Review

Eddy AI does not guarantee that:

  • alerts will be delivered instantly
  • alerts will reach the intended recipient without delay
  • alerts will be seen, reviewed, or acted upon in a timely manner

Responsibility for monitoring notifications and ensuring appropriate response sits entirely with the supervising organisation.

4.5 No Guarantee of Escalation or Intervention

Eddy AI does not:

  • escalate concerns to external agencies
  • initiate safeguarding procedures
  • ensure that any identified issue is acted upon

There is no guarantee that any risk, behaviour, or safeguarding concern will be:

  • detected
  • flagged
  • reviewed
  • escalated
  • resolved

4.6 Responsibility of Supervising Parties

All supervising organisations, parents, guardians, and responsible individuals must:

  • maintain appropriate human oversight
  • implement their own monitoring and safeguarding processes
  • not rely on Eddy AI as a sole or primary safeguard

Eddy AI is a supplementary tool and must be used alongside, not in place of, established safeguarding systems.

4.7 Acceptance of System Limitations

By using Eddy AI, all supervising parties acknowledge and accept that:

  • the system does not provide monitoring or supervision
  • alerts are not guaranteed
  • reliance on the system alone is inappropriate
  • full responsibility for awareness, response, and safeguarding remains with them

5. Responsibility for Safeguarding and Welfare

Full responsibility for safeguarding, duty of care, and user welfare remains at all times with the supervising individual or organisation.

Use of Eddy AI does not transfer, reduce, or share this responsibility.

5.1 Primary Responsibility

The supervising organisation or responsible individual retains sole responsibility for:

  • the safety, wellbeing, and protection of all users
  • all safeguarding decisions and actions
  • compliance with applicable legal, regulatory, and statutory frameworks

This applies at all times, regardless of whether Eddy AI is in use.

5.2 Monitoring and Oversight Responsibilities

Supervising organisations must:

  • actively monitor user wellbeing, behaviour, and engagement
  • maintain appropriate levels of human supervision
  • ensure that users are supported through established pastoral or safeguarding systems

Eddy AI does not provide monitoring and must not be relied upon for this purpose.

5.3 Review of Alerts and System Outputs

Supervising organisations are responsible for:

  • reviewing all alerts generated by Eddy AI
  • ensuring alerts are monitored through appropriate internal processes
  • interpreting alerts using professional judgment and contextual understanding

Alerts must be treated as assistive signals only and not as conclusions.

5.4 Decision Making and Safeguarding Action

All decisions relating to safeguarding must be made by the supervising organisation, including:

  • whether a concern requires further investigation
  • whether intervention or support is required
  • whether escalation to internal or external parties is necessary

Eddy AI does not make, influence, or determine safeguarding decisions.

5.5 Escalation and Intervention

The supervising organisation is solely responsible for:

  • initiating safeguarding procedures
  • escalating concerns in line with internal policies
  • engaging designated safeguarding leads or equivalent roles
  • contacting external agencies where required

Eddy AI does not initiate or carry out escalation.

5.6 Legal and Regulatory Compliance

Supervising organisations must:

  • comply with all applicable safeguarding laws and statutory guidance
  • meet sector specific obligations
  • maintain appropriate records, reporting, and audit trails

Use of Eddy AI does not alter or satisfy these obligations.

5.7 No Transfer or Sharing of Duty of Care

Use of Eddy AI does not:

  • transfer duty of care toAmbertrace Labs
  • create any shared safeguarding responsibility
  • establish any expectation of monitoring or intervention byAmbertrace

All duty of care remains solely with the supervising organisation or individual.

5.8 Role of Professional Judgment

All outputs generated by Eddy AI must:

  • be reviewed by appropriately trained individuals
  • be considered alongside broader contextual information
  • not be relied upon in isolation

Professional judgment must always take precedence over automated outputs.

5.9 Integration with Organisational Safeguarding Systems

Organisations must ensure that:

  • Eddy AI is used alongside existing safeguarding frameworks
  • alerts are incorporated into established workflows where appropriate
  • system use aligns with internal policies and statutory requirements

Eddy AI is a supporting tool, not a safeguarding system.

5.10 Accountability for Actions and Outcomes

The supervising organisation retains full responsibility for:

  • how alerts are interpreted
  • what actions are taken
  • the timing and appropriateness of interventions
  • the outcomes of safeguarding decisions

Ambertrace Labshas no responsibility for these actions or outcomes.

5.11 Acknowledgement of Responsibility

By using Eddy AI, supervising organisations, parents, guardians, and oversight users acknowledge and confirm that:

  • they retain full safeguarding responsibility at all times
  • they will not rely on the system as a primary safeguard
  • they will exercise independent professional judgment
  • they accept that Eddy AI does not replace safeguarding processes or duties

6. No Clinical or Professional Advice

Eddy AI does not provide clinical, medical, therapeutic, or professional advice of any kind.

All outputs generated by the system are informational and supportive only and must not be relied upon as a substitute for qualified professional judgment.

6.1 No Medical or Mental Health Advice

Eddy AI does not:

  • provide medical advice
  • provide mental health advice
  • assess physical or psychological conditions
  • recommend treatment or clinical intervention

The system is not a healthcare service and must not be used for medical purposes.

6.2 No Diagnosis or Assessment

Eddy AI does not:

  • diagnose mental health conditions
  • perform psychological or behavioural assessments
  • determine levels of risk, severity, or need
  • produce clinically validated evaluations

Any outputs must not be interpreted as a diagnosis or formal assessment.

6.3 No Therapy or Counselling

Eddy AI does not:

  • provide therapy, counselling, or therapeutic intervention
  • deliver structured treatment or care plans
  • replace interaction with qualified professionals
  • establish a therapeutic relationship with users The system is not a substitute for trained practitioners.

6.4 No Crisis Intervention

Eddy AI does not:

  • identify or manage crisis situations with certainty
  • provide real time intervention or support
  • respond to emergencies or urgent needs
  • offer crisis management or safeguarding intervention

The platform must not be relied upon in situations requiring immediate or specialist support.

6.5 Nature of Outputs

All outputs generated by Eddy AI, including:

  • alerts
  • responses
  • prompts
  • suggestions

are:

  • general in nature
  • generated through automated processes
  • non personalised in a clinical or professional sense

Outputs are intended to support awareness and engagement only.

6.6 Requirement for Professional Judgment

All outputs must:

  • be reviewed by appropriately trained individuals
  • be considered alongside full contextual information
  • not be relied upon in isolation

Decisions relating to wellbeing, safeguarding, or intervention must be made by qualified professionals.

6.7 Prohibition on Reliance

Users and supervising organisations must not:

  • rely on Eddy AI as a source of professional advice
  • treat outputs as authoritative or definitive
  • delay seeking professional support based on system outputs
  • use the system in place of appropriate services

6.8 Responsibility for Seeking Professional Support

Where required, it is the responsibility of:

  • the supervising organisation
  • parents or guardians
  • responsible individuals

to seek appropriate support from:

  • healthcare professionals
  • mental health services
  • safeguarding authorities
  • other qualified specialists

6.9 Acknowledgement of Limitations

By using Eddy AI, all users and supervising parties acknowledge and accept that:

  • the system does not provide professional advice
  • outputs are informational only
  • reliance on the system for clinical or safeguarding decisions is inappropriate
  • responsibility for decisions and actions remains with them

7. Emergency Situations

In the event of immediate risk, danger, or crisis, users, supervising organisations, parents, guardians, and responsible individuals must take direct and immediate action outside of the Eddy AI platform.

Eddy AI is not an emergency service and must not be used, treated, or relied upon as an emergency response system.

7.1 No Emergency Response Function

Eddy AI does not:

  • provide emergency assistance
  • provide immediate human intervention
  • contact emergency services
  • contact safeguarding authorities automatically
  • guarantee urgent detection of crisis situations
  • provide live crisis management or support

The platform has no operational ability to respond to emergencies in real time.

7.2 Situations Requiring Immediate Action

Immediate action must be taken where there is any indication of:

  • risk of serious harm to self or others
  • suicidal statements or threats
  • self harm disclosures requiring urgent attention
  • abuse, exploitation, or immediate safeguarding danger
  • serious mental or emotional distress requiring urgent support
  • medical emergency or physical danger
  • any situation in which delay could increase the risk of harm

In such circumstances, reliance must not be placed on Eddy AI to identify, assess, confirm, or manage the situation.

7.3 Required Response by Users and Supervising Parties

In any emergency or crisis situation, the responsible person or organisation must:

  • contact emergency services where urgent intervention is required
  • follow internal safeguarding and escalation procedures immediately
  • notify the designated safeguarding lead or equivalent responsible professional
  • engage appropriate support services, agencies, or healthcare providers without delay
  • take any other protective action required by law, policy, or professional judgment

These steps must be taken independently of any alert, output, or response generated by Eddy AI.

7.4 No Reliance on Alerts or Absence of Alerts

Users and supervising parties must not:

  • wait for an alert before taking action
  • delay intervention while checking the platform
  • treat absence of an alert as evidence that no emergency exists
  • rely on system outputs to decide whether a situation is urgent

The absence of an alert does not mean that a user is safe or that no crisis is present.

7.5 No Guarantee of Detection or Timing

Eddy AI does not guarantee that:

  • emergency situations will be detected
  • urgent risk will be correctly identified
  • alerts will be generated in time
  • alerts will reflect the seriousness of the situation
  • notifications will be seen or acted upon immediately

There may be gaps, delays, inaccuracies, or failures in detection and response.

7.6 Responsibility for Emergency Action

Full responsibility for identifying and responding to emergency situations remains at all times with:

  • the supervising organisation
  • parents or guardians
  • responsible staff or professionals
  • any individual with oversight or duty of care

Ambertrace Labsdoes not assume responsibility for emergency assessment, emergency escalation, or emergency intervention.

7.7 Relationship with Safeguarding Duties

Nothing in Eddy AI replaces or reduces the obligation of supervising organisations or responsible individuals to:

  • act immediately where risk is suspected
  • comply with safeguarding law and statutory guidance
  • follow internal emergency and safeguarding procedures
  • use professional judgment in urgent situations

Eddy AI is a supplementary tool only and must never be used as a substitute for direct action.

7.8 Acknowledgement of Emergency Limitations

By using Eddy AI, organisations and supervising parties acknowledge and accept that:

  • Eddy AI is not an emergency response system
  • it must not be relied upon in situations of immediate risk or danger
  • urgent concerns must be handled outside the platform through direct action
  • responsibility for emergency response remains entirely with them

If you want, I can now do the next section in the same format so the whole policy reads consistently.

8. System Configuration and Control

Supervising organisations are solely responsible for configuring, managing, and maintaining the Eddy AI alert system within their operational environment.

Eddy AI provides configurable functionality, but all decisions regarding how the system is set up, used, monitored, and acted upon remain with the supervising organisation.

8.1 Organisational Control of Configuration

The supervising organisation is responsible for all configuration decisions, including:

  • enabling or disabling alert functionality
  • selecting which detection rules or triggers are active
  • configuring thresholds, sensitivities, or parameters where available
  • determining which features are used within their environment

These decisions must be made in line with the organisation’s safeguarding framework, operational needs, and legal obligations.

8.2 Responsibility for Notification Settings

Supervising organisations must determine:

  • which individuals or roles receive alerts
  • how alerts are distributed across teams or departments
  • how frequently notifications are received
  • how alerts are prioritised and managed internally The organisation is responsible for ensuring that notification settings are appropriate, effective, and actively monitored.

8.3 Assignment of Responsible Individuals

Organisations must ensure that:

  • clear responsibility is assigned for reviewing alerts
  • designated staff or safeguarding leads are identified as recipients
  • roles and responsibilities are documented and understood
  • sufficient coverage exists to ensure alerts are not missed

Eddy AI does not assign responsibility for alert handling.

8.4 Ongoing Management and Oversight

Supervising organisations are responsible for:

  • regularly reviewing and updating system configurations
  • ensuring settings remain appropriate as risks or requirements change
  • monitoring the effectiveness of alert rules and thresholds
  • ensuring alerts are integrated into operational workflows

Configuration must not be treated as a one time setup and must be actively maintained.

8.5 Alignment with Safeguarding Procedures

Organisations must ensure that:

  • system configuration aligns with internal safeguarding policies
  • alert thresholds and rules reflect organisational risk tolerance
  • alert handling is embedded within existing safeguarding processes
  • system use supports, rather than replaces, established procedures

Eddy AI must be configured as part of a wider safeguarding framework.

8.6 No Control by AmbertracePost Deployment

Once deployed, Ambertrace Labs:

  • does not control how alerts are configured
  • does not manage or adjust organisational settings
  • does not determine who receives notifications
  • does not monitor whether alerts are reviewed or acted upon

All operational control sits entirely with the supervising organisation.

8.7 No Responsibility for Configuration Outcomes

Ambertrace Labsis not responsible for:

  • how the system is configured by the organisation
  • any gaps or issues arising from configuration choices
  • missed alerts due to disabled features or incorrect settings
  • ineffective or inappropriate use of alert rules

All outcomes arising from configuration decisions remain the responsibility of the supervising organisation.

8.8 Requirement for Competent Configuration

Supervising organisations must ensure that:

  • system configuration is carried out by competent and authorised individuals
  • staff responsible for configuration understand the system’s capabilities and limitations
  • appropriate governance and oversight is applied to configuration decisions

Improper configuration may impact system effectiveness and remains the responsibility of the organisation.

8.9 Acknowledgement of Configuration Responsibility

By using Eddy AI, supervising organisations acknowledge and confirm that:

  • they are fully responsible for configuring and managing the system
  • they control how alerts are generated, distributed, and handled
  • Ambertrace Labshas no control over operational use once deployed
  • all risks associated with configuration and usage remain with them

9. Limitations of AI Systems

AI based systems are inherently limited and operate on probabilistic models, pattern recognition, and predefined logic rather than human understanding or judgment.

Eddy AI is designed to provide assistive signals only and must not be relied upon as a definitive or complete system for identifying risk, behaviour, or safeguarding concerns.

9.1 Inherent Limitations of AI

Eddy AI operates using automated processes that:

  • analyse patterns in data and user interactions
  • apply rules, thresholds, and classification models
  • generate outputs based on probability, not certainty

As a result, the system does not possess:

  • human understanding or reasoning
  • full contextual awareness
  • the ability to verify accuracy of user inputs
  • the ability to assess intent, truthfulness, or real world circumstances

9.2 Risk of False Positives

The system may:

  • flag content or behaviour that does not represent a genuine concern
  • misclassify neutral or benign interactions as risk indicators
  • generate alerts based on incomplete or ambiguous data

False positives may result in unnecessary alerts and must be assessed by a human.

9.3 Risk of False Negatives

The system may:

  • fail to identify genuine safeguarding or wellbeing concerns
  • not trigger alerts where risk is present
  • miss subtle, complex, or context dependent indicators

Absence of an alert does not indicate absence of risk.

9.4 Contextual Limitations

Eddy AI may:

  • misinterpret tone, language, or intent
  • fail to understand cultural, social, or situational context
  • incorrectly interpret sarcasm, humour, or indirect communication
  • lack awareness of offline circumstances affecting the user

All outputs must therefore be interpreted with caution.

9.5 Dependence on Input Quality

System outputs are dependent on the quality and completeness of input data.

Where data is:

  • incomplete
  • inaccurate
  • misleading
  • not representative of real circumstances

outputs may be unreliable or incorrect.

9.6 Variability of Outputs

Outputs generated by Eddy AI may:

  • vary depending on phrasing, timing, or context of input
  • change over time as interactions evolve
  • produce inconsistent results in similar situations The system does not guarantee consistency or repeatability of outputs.

9.7 Limitations in Detection and Timing

Eddy AI does not guarantee:

  • detection of all relevant behaviours or risks
  • timely identification of emerging concerns
  • alignment between alert severity and real world risk

There may be delays, gaps, or inaccuracies in detection.

9.8 No Replacement for Human Judgment

Due to these limitations:

  • all outputs must be reviewed by appropriately trained individuals
  • decisions must be based on professional judgment and full context
  • the system must not be relied upon in isolation

Human oversight is essential at all times.

9.9 No Guarantee of System Effectiveness

Eddy AI does not guarantee:

  • accuracy of outputs
  • completeness of detection
  • suitability for all users or environments
  • effectiveness in preventing harm or identifying risk

The system is a supportive tool only.

9.10 Acknowledgement of Limitations

By using Eddy AI, all supervising organisations, parents, guardians, and users acknowledge and accept that:

  • AI systems are inherently limited and probabilistic
  • false positives and false negatives may occur
  • outputs may be inaccurate, incomplete, or misinterpreted
  • reliance on the system alone is inappropriate
  • full responsibility for interpretation, decision making, and action remains with them

10. Limitation of Liability

To the fullest extent permitted by applicable law, the following limitations of liability apply to the use of Eddy AI and its associated features, including the wellbeing alerts and escalation system.

Eddy AI is provided as an assistive technology tool only and does not assume responsibility for safeguarding, monitoring, or intervention.

10.1 No Responsibility for Detection or Escalation

Ambertrace Labsand its affiliates accept no responsibility for:

  • any failure to detect, identify, or flag a concern
  • any failure to generate an alert
  • any failure to escalate a safeguarding issue
  • any delay in detection, alerting, or escalation

The system does not guarantee that risks or concerns will be identified or acted upon.

10.2 No Responsibility for Interpretation or Use

Ambertrace Labsand its affiliates accept no responsibility for:

  • how alerts, outputs, or system generated information are interpreted
  • decisions made based on system outputs
  • actions or inactions taken in response to alerts
  • misuse or inappropriate use of the system

All interpretation and use of outputs is the responsibility of the supervising organisation.

10.3 No Liability for Harm, Loss, or Damage

Ambertrace Labsand its affiliates shall not be liable for:

  • any harm, injury, or damage to any individual
  • any safeguarding failure or incident
  • any loss arising from reliance on the system
  • any indirect, consequential, or incidental loss or damage

This includes, but is not limited to, harm arising from:

  • missed or inaccurate alerts
  • delayed or failed responses
  • incorrect interpretation of outputs

10.4 No Duty of Care Assumed

Use of Eddy AI does not:

  • create any duty of care on the part ofAmbertrace Labs
  • establish any monitoring or supervisory obligation
  • transfer or share safeguarding responsibility

Ambertrace Labsdoes not assume responsibility for user welfare at any time.

10.5 Responsibility of Supervising Organisations

All responsibility for safeguarding decisions and actions remains with the supervising individual or organisation.

This includes responsibility for:

  • monitoring users
  • reviewing alerts
  • making safeguarding decisions
  • initiating escalation or intervention
  • complying with legal and regulatory obligations

10.6 Configuration and Operational Responsibility

Ambertrace Labsand its affiliates are not responsible for:

  • how the system is configured by the organisation
  • who receives alerts or how they are managed
  • whether alerts are monitored or acted upon
  • any outcomes resulting from configuration choices

All operational responsibility remains with the supervising organisation.

10.7 Use at Own Risk

Use of Eddy AI is at the sole risk of the supervising organisation.

Organisations must ensure that:

  • the system is used appropriately and within its intended scope
  • reliance is not placed solely on automated outputs
  • independent safeguarding processes remain in place

10.8 No Liability for Third Party Actions

Ambertrace Labsand its affiliates are not responsible for:

  • actions or omissions of supervising organisations, staff, or users
  • decisions made by third parties based on system outputs
  • failures in organisational safeguarding processes

10.9 No Limitation Where Prohibited by Law

Nothing in this policy excludes or limits liability where such limitation is not permitted under applicable law, including liability for:

  • death or personal injury caused by negligence
  • fraud or fraudulent misrepresentation
  • any other liability which cannot lawfully be excluded

10.10 Acknowledgement and Acceptance

By using Eddy AI, all organisations, supervising parties, and users acknowledge and accept that:

  • the system does not guarantee detection, accuracy, or response
  • Ambertrace Labsis not responsible for safeguarding outcomes
  • reliance on the system is at their own risk
  • full responsibility for all decisions, actions, and consequences remains with them

11. Acceptance of Terms

By accessing, deploying, enabling, or using Eddy AI, all supervising organisations, parents, guardians, and authorised users confirm that they have read, understood, and agreed to the terms set out in this policy.

Use of Eddy AI constitutes acceptance of the system’s purpose, limitations, and the allocation of responsibilities described throughout this document.

11.1 Confirmation of Understanding

Supervising organisations and overseeing parties confirm that they:

  • understand that Eddy AI is an assistive technology tool only
  • recognise that outputs are automated, probabilistic, and non definitive
  • understand that the system does not provide monitoring, safeguarding, or intervention

11.2 Acceptance of Safeguarding Responsibility

Supervising organisations and responsible individuals confirm that they:

  • retain full responsibility for safeguarding, duty of care, and user welfare
  • are responsible for monitoring, reviewing, and responding to alerts
  • will make all safeguarding decisions using independent professional judgment
  • will maintain appropriate safeguarding procedures and oversight

11.3 Acceptance of System Limitations

All parties acknowledge and accept that:

  • the system may fail to detect risks or concerns
  • alerts may be inaccurate, incomplete, or delayed
  • outputs must not be relied upon in isolation
  • gaps in detection and response may exist

11.4 Acceptance of Limitation of Liability

Supervising organisations, parents, guardians, and users confirm that they:

  • accept the limitation of liability provisions set out in this policy
  • understand that Ambertrace Labsdoes not assume safeguarding responsibility
  • acknowledge that Ambertrace Labsis not liable for outcomes arising from use of the system

11.5 Appropriate Use and Reliance

By using Eddy AI, supervising organisations agree that they will:

  • use the system in accordance with this policy and applicable law
  • not rely on Eddy AI as a primary safeguarding mechanism
  • not treat alerts or outputs as definitive conclusions
  • not delay action while awaiting system outputs

11.6 Binding Effect of Use

Use of Eddy AI constitutes:

  • agreement to the allocation of responsibility set out in this policy
  • acceptance that all safeguarding responsibility remains with the supervising organisation
  • acknowledgement that Eddy AI is a supplementary tool only

11.7 Responsibility for Communication

Supervising organisations are responsible for:

  • ensuring that relevant staff understand and comply with this policy
  • communicating appropriate expectations to users, parents, and guardians where applicable
  • ensuring that the system is used in a manner consistent with safeguarding and data protection obligations

11.8 Ongoing Acceptance

Continued use of Eddy AI confirms ongoing acceptance of this policy.

Organisations must ensure that:

  • use of the system remains compliant with internal policies and legal requirements
  • staff continue to act within their responsibilities
  • any updates to this policy are reviewed and implemented appropriately

11.9 Acknowledgement of Responsibility

By using Eddy AI, all supervising organisations, parents, guardians, and oversight users acknowledge and confirm that:

  • they have understood the nature and limitations of the system
  • they retain full responsibility for safeguarding and welfare
  • they accept that Eddy AI does not assume any duty of care
  • they accept all risks associated with use of the system within the limits of applicable law