Our approach to Artificial Intelligence.

Gempro may use AI-enabled tools to support activities such as:

• research and information gathering
• brainstorming and concept development
• copywriting and editing
• design exploration and creative ideation
• creating or refining visual concepts
• data analysis and pattern identification
• summarising information
• administrative and repetitive tasks
• improving internal workflows and productivity
• software, web and digital development
• quality assurance and checking
• helping our team explore different solutions to a customer brief

AI may help us get to a stronger answer faster. It does not remove the responsibility of the Gempro team member delivering the work.

We do not regard AI output as inherently correct simply because it has been generated by an advanced system.

Where AI contributes to customer work, a Gempro team member remains responsible for reviewing the output and determining whether it is appropriate to use. Depending on the nature of the work, this may include checking:

• factual accuracy
• context and relevance
• brand accuracy
• spelling and language
• calculations
• sources
• bias or misleading assumptions
• intellectual property considerations
• confidentiality
• image accuracy
• suitability for the intended audience

Where accuracy is particularly important, AI-generated information should be verified using appropriate authoritative sources rather than relying solely on the AI response.

We believe there is an important difference between enhancement and deception. Gempro will not knowingly use AI to:

• fabricate customer testimonials
• create false case studies
• invent customers, partnerships or credentials
• impersonate a real person without appropriate authority
• fabricate quotations attributed to real individuals
• deliberately create misleading representations of products or services
• conceal significant mistakes behind AI-generated content
• knowingly present AI-generated information as verified fact when it has not been appropriately checked

Where synthetic imagery, video, audio or another AI-generated asset could reasonably cause somebody to believe something occurred in the real world when it did not, we will consider appropriate disclosure or labelling.

Customer trust is fundamental to Gempro. We therefore treat the use of customer information within AI systems with appropriate care.

Personal data must be handled consistently with applicable UK data protection requirements. The UK Information Commissioner's Office provides specific guidance regarding AI systems that process personal information, including requirements relating to lawfulness, fairness, transparency, accountability and risk management.

Gempro personnel should not place confidential, commercially sensitive or personal customer information into public or unapproved AI systems where doing so could expose that information or use it inappropriately. Where AI involves the processing of personal information, we consider factors including:

• the purpose for using the information
• whether its use is lawful, fair and transparent
• the minimum information necessary
• security and access
• retention
• potential risks to individuals
• whether additional assessment or safeguards are necessary

The ICO recommends a risk-based approach to AI and provides tools for assessing AI-related risks to people's rights and freedoms.

A customer giving Gempro information does not automatically give us permission to make that information publicly available or use it for unrelated purposes.

Customer information, documents, designs, artwork, strategies and commercially sensitive materials should only be used for their intended purpose and handled in accordance with our contractual, confidentiality and data-protection obligations.

We take particular care when entering customer materials into third-party AI platforms.

Generative AI creates new considerations around copyright, ownership and originality. We therefore expect our team to exercise appropriate judgement when AI contributes to creative work.

We will not knowingly instruct AI to reproduce copyrighted work unlawfully or deliberately pass another organisation's intellectual property off as our own. Where originality, licensing or ownership is commercially important, we may apply additional checks or use appropriately licensed assets and tools.

AI is part of the creative toolkit. It is not a shortcut around intellectual property rights.

Gempro may use generative AI in appropriate circumstances to create or develop imagery, concepts, layouts, copy, video, audio or other creative materials.

Our question is not simply "Was AI used?" Our questions are: "Is it good? Is it accurate? Is it appropriate? Is it ethical? And does it honestly represent what we are communicating?"

When AI-generated or AI-manipulated media could materially mislead someone, we believe transparency is important.

We only build an AI presenter with written consent from the person filmed, and we use it only for the organisation that commissioned it.

Avatar Studio videos carry an AI-generated label as standard, so viewers always know.

If a licence ends, or the presenter leaves the business, we delete the presenter model and voice within 30 days.

Gempro does not believe important decisions about people should be casually delegated to an algorithm.

We will not knowingly rely solely on generative AI to make significant decisions concerning an individual's employment, capability, suitability or other similarly consequential matters. AI may provide information or assistance, but appropriate human judgement and accountability must remain part of significant decisions.

This approach also aligns with the ICO's emphasis on transparency, explainability, fairness and accountability when AI is used in ways affecting individuals.

Generative AI can produce information that sounds convincing but is incorrect. For that reason, Gempro does not treat AI output as evidence.

Where factual accuracy matters, particularly in technical, financial, legal, medical, regulatory or safety-related material, information should be appropriately reviewed and verified before being presented as fact.

We will correct material AI-assisted errors when we become aware of them.

AI systems can reflect or amplify biases contained within their training data or instructions. We expect our people to consider the potential for unfair, discriminatory or inappropriate output when using AI.

Human review is particularly important where AI-generated content relates to people, cultures, communities, accessibility or other potentially sensitive contexts.

Fairness is also an established principle within UK data protection expectations for AI systems processing personal information.

Not every AI tool should automatically be trusted with Gempro or customer information. Before adopting AI technologies for significant business use, we may consider factors such as:

• security
• privacy
• contractual terms
• how submitted data is processed
• whether submitted information may be used for model training
• data retention
• intellectual property terms
• access controls
• reliability
• the sensitivity of the intended use

The level of assessment should be proportionate to the risk involved. A tool used to brainstorm a headline does not require the same level of scrutiny as a system processing customer or personal data.

A human remains accountable for Gempro's work.

AI cannot take responsibility for a decision, defend a piece of advice to a customer or understand a relationship in the same way a person can. Our people therefore remain responsible for the work they produce, regardless of the technology used to help create it.

We do not believe responsible AI requires putting an "AI generated" label on every spellcheck, recommendation or productivity enhancement.

We do believe customers should be able to understand when AI has played a material role in something where that fact could influence their interpretation, trust or decision.

If a customer has particular contractual, regulatory or confidentiality requirements concerning AI, we expect those requirements to be agreed and followed. And if a customer wants to understand how AI has been used within their project, we will have that conversation openly.

AI technology is developing extremely quickly. Our approach will therefore evolve as technology, customer expectations, legislation, regulation and industry best practice develop.

The ICO notes that its detailed AI and data-protection guidance is currently under review following changes introduced by the Data (Use and Access) Act, illustrating why this policy needs periodic review rather than being treated as static.

Gempro will periodically review this policy and our use of AI to make sure our approach remains responsible, practical and appropriate.