‍Our Ethos

‍

About Sign Languages

Sign languages are real languages.

They have their own grammar, structure, and culture.

They should be treated with the same respect as spoken languages.

Barriers for Deaf People

Many Deaf people still face barriers.

• There are not enough sign language interpreters.
• Public information is not always available in sign language.
• Many websites and digital services are not accessible.

How Artificial Intelligence Can Help

Artificial Intelligence, often called AI, is becoming more powerful.

AI can help translate languages and share information.

At Signapse, we believe AI can help make the world more accessible for Deaf people.

AI must be developed carefully, ethically, and with Deaf communities.

Our Core Principles

Signapse follows important principles when developing AI for sign language.

Principle 1: Sign Languages Are Full Languages

Sign languages are complete languages with grammar and cultural importance.

Sign language users have linguistic rights that must be respected.

Principle 2: AI Should Increase Access

AI should increase accessibility but should not replace human interpreters in sensitive situations.

• Healthcare
• Legal settings
• Child protection

Principle 3: Work With Deaf Communities

AI must be developed together with Deaf people.

Deaf people should be involved in design, testing, governance, and deployment.

Principle 4: Ethical Use of Data

Sign language data must respect privacy, consent, ownership, and fair payment.

People should control how their signing and digital likeness are used.

Principle 5: Protect Sign Language Culture

AI must support sign languages and preserve grammar, meaning, and dialects.

Principle 6: Transparency and Oversight

People should know when AI is used.

Signapse systems allow review, corrections, and human oversight.

Principle 7: Inclusive AI

AI should work for all sign language users.

Datasets should include diverse signers and regional variations to reduce bias.

Our Vision

Signapse wants a future where sign language access is everywhere.

• Public services are accessible in sign language.
• Education and employment are accessible.
• Digital platforms support sign language.
• Sign languages are protected and respected.

---------------------------------------------------------------------------------------------------------------

Content Moderation Policy

‍

This document explains how Signapse checks content.

Content means videos or media used in our services.

This includes videos uploaded by customers and videos translated into sign language.

Our goal is to keep communication accessible, safe, and respectful.

‍

Our principles

‍

We follow some important principles when moderating content.

• We follow the law.

• We protect safety and inclusion.

• We explain our rules clearly.

• We combine technology and human review.

• We make moderation accessible for Deaf users.

‍

Content that is not allowed

‍

Some content is not allowed on Signapse services.

• Violent extremism or hate groups.

• Hate speech against protected groups.

• Child sexual abuse material.

• Pornography created for sexual arousal.

• Encouraging violence or self-harm.

• Fraud, scams, or impersonating authorities.

‍

Content that may need extra checks

‍

Some content may be allowed but needs extra review.

• Medical advice.

• Legal advice.

• Financial advice.

• Political content.

• Sensitive social topics.

‍

How content is checked

‍

• A client uploads a video.

• The system checks the video automatically.

• If a problem is detected, a human moderator reviews the content.

• If approved, the content is translated into sign language.

• The translated video may also be checked again.

‍

Reporting a problem

‍

Users can report harmful or inappropriate content.

• Reports can be sent to support@signapse.com.

• The Customer Success Team reviews reports within 48 hours.

‍

Appeals

‍

If a user disagrees with a moderation decision, they can appeal.

• Appeals are normally reviewed within 5 working days.

• The appeal will be reviewed by someone different from the first moderator.

‍

Privacy and data protection

‍

• We follow our data protection policy.

• We collect only minimal personal data.

• Flagged content is handled securely.

‍

Who is responsible

‍

• The CEO owns this policy.

• The CTO manages moderation technology.

• The Head of Customer Success manages moderation and user reports.

‍

Our commitment

‍

Signapse is a small company but we take moderation seriously.

We will improve our moderation systems as the company grows.

Our goal is to protect users and build trust.

‍

---------------------------------------------------------------------------------------------------------------

AI & Human Translation Use Policy

‍

1. Why This Policy Exists

This policy explains when Signapse recommends AI sign-language translation and when we recommend a qualified human translator or interpreter.

Our goal is simple: use AI where it is safe and appropriate, and use humans where the stakes are high.

This applies to all Signapse products: SignStudio, SignStream, plugins and APIs.

We believe sign languages are full natural languages and Deaf communities are rights-holders.

We use guardrails. They are not walls,  they guide safe innovation.

2. Core Principles

2.1 Safety First

AI may only be used where it is accurate enough and does not create risk of harm or misunderstanding.

2.2 Human First for High-Stakes Situations

If communication affects health, legal rights, liberty, safeguarding or serious employment consequences, a qualified interpreter is required.

2.3 AI for Speed and Scale

AI works well for routine, repeatable, non-critical informational content.

2.4 Transparency and Oversight

AI-generated content is labelled where possible. All deployments include monitoring and escalation pathways.

2.5 Consent and Representation

AI output must not imply endorsement or identity of a real signer. Where recorded human signers are used, informed consent and compensation apply.

3. When to Use AI vs Human Translation

Transport

Use a Human Interpreter for:

• Live two-way passenger interaction
• Safeguarding situations
• Complaints or disputes

AI may be used for:

• Routine service announcements
• Journey updates
• Wayfinding information
• Websites and digital displays

Community

Use a Human Interpreter for:

• Sensitive community issues
• Safeguarding communication
• Emergency instructions involving risk to life

AI may be used for:

• Council announcements
• Local service updates
• Cultural content
• General public campaigns
• Website content

Employment

Use a Human Interpreter for:

• Contract negotiation
• Grievances or disputes
• Performance management

AI may be used for:

• Routine team updates
• Internal training
• Standard operating procedures

Education

Use a Human Interpreter for:

• Serious disciplinary meetings
• Individual Education Plans (IEPs)
• Safeguarding conversations
• Complex academic content

AI may be used for:

• Pre-recorded curriculum videos (quality reviewed)
• Routine announcements
• Parent newsletters
• Website content

Health

Always use a Human Interpreter for:

• Live consultations
• Diagnosis and treatment discussions
• Mental health consultations
• Safeguarding conversations

AI may be used for:

• General public health information
• Preventative campaigns
• Administrative updates
• Appointment reminders

Legal

Always use a Human Interpreter for:

• Live legal communication
• Contracts and legal advice
• Police or court settings
• Anything affecting liberty or finances

AI may be used for:

• General awareness videos
• Non-binding informational content

Safeguarding

Always human. No exceptions.

4. Escalation Path

• Level 1 – Customer Support (48 hours)
• Level 2 – Moderation Panel (3 working days)
• Level 3 – CEO (5 working days)

5. Policy Review

This policy is reviewed every 6 months. Major updates create a new version. A sign-language video summary accompanies this policy.

‍

---------------------------------------------------------------------------------------------------------------

Signapse Translation Quality Assurance Policy

‍

This document explains how Signapse checks the quality of sign language translations.

Translations may be created using Artificial Intelligence (AI), human translators, or Signapse tools.

This policy applies to all translations produced by Signapse.

‍

Our principles

‍

Signapse follows important principles to make sure translations are high quality.

• Accuracy – translations must communicate the correct meaning.

• Clarity – translations must be easy for Deaf sign language users to understand.

• Relevance – the tone and style should match the situation.

• Consistency – terminology and signs should stay consistent.

• Risk awareness – higher‑risk sectors need additional checks.

• Continuous improvement – we improve our systems using feedback.

• Transparency – we tell users when translations are generated by AI.

‍

Laws and regulations

‍

Signapse follows important laws and standards.

• Artificial Intelligence Act (EU).

• UK Equality Act 2010.

• European Accessibility Act

• Data protection laws such as GDPR.

‍

How translations are checked

‍

• Step 1 – Identify the content, audience, sector and risk level.

• Step 2 – Decide whether AI translation is appropriate.

• Step 3 – Generate the translation using AI or human translators.

• Step 4 – Review the translation for accuracy, clarity and cultural meaning.

• Step 5 – Deliver the translation and label it as AI‑generated if appropriate.

• Step 6 – Collect feedback from users and clients.

• Step 7 – Use feedback to improve systems and processes.

‍

Internal quality checks

‍

• Translation workflow guides.

• Glossaries and style guides.

• Quality assurance checklists.

• Review by internal sign language specialists.

‍

External feedback

‍

• Deaf user testing groups.

• Consultation with Deaf organisations.

• User feedback from real projects and trials.

‍

Who is responsible

‍

• CEO – responsible for overall quality strategy.

• CTO – responsible for translation technology and AI systems.

• Customer Success – responsible for research tools and AI quality.

• Head of Quality – responsible for quality checklists and reviews.

• Chief Revenue Officer – responsible for client feedback.

• Human translators and specialists support quality review.

‍

Quality measures

‍

• SignStream 5‑Star rating system.

• AI translation accuracy targets.

• Turnaround time for translations.

• Monitoring translation errors.

• User satisfaction levels.

‍

Policy review

‍

• This policy will be reviewed every six months.

• Internal audits will take place regularly.

• External reviews with the Deaf community will help ensure quality.

‍

Sign‑off

‍

Sally Chalk

CEO, Signapse Ltd.

Date: 9 February 2026

‍

‍

‍