Policy 07
AI & Responsible Technology
“AI can assist the work. It doesn’t own the judgement.”
We use technology, including generative AI, to help solve communication problems. This policy makes sure experimentation never outruns responsibility, and that a human remains accountable for everything we deliver.
At a glance
The short version.
Human review, always
AI output is treated as potentially wrong, and facts, visuals and technical details are checked before anything reaches a client.
Your data stays protected
Confidential or personal client material is never uploaded to an AI tool that hasn’t been approved for it.
Open with clients
We discuss material AI use where it affects rights, confidentiality or reliability, and respect any client who says no to AI.
In practice
What we commit to.
- Never pass unreviewed generated content to a client because of a deadline
- Never train or fine-tune a model on client assets outside the agreed scope
- Consider licensing, provenance and third-party rights in AI output
- Never create deceptive synthetic media or clone a voice without proper authority
- Don’t use monitoring or scoring tools just because they exist
- Check new tools for data use, retention and security before they touch sensitive material
- Change method if fixing AI errors takes longer than doing the work properly
The detail
Everything else.
The key parts of the policy, in plain English. The downloadable PDF is the full, controlled version.
What we may use it for
Brainstorming, research (verified), drafting non-sensitive text, technical and coding help, image and video ideation and previs, checkable production tasks, transcription and captioning with review, and workflow automation that doesn’t make inappropriate decisions about people.
Personal data and automated decisions
Our privacy policy applies to any AI processing of personal data. We don’t rely on solely automated decisions with significant effects on people unless lawful and specifically approved, and human review must be meaningful.
Likeness and voice
Extra care applies to children, public-interest content, political material, testimonials and factual communication.
Respectful technology
Technology should reduce friction, not reduce people to metrics. Staff are told about any material workplace monitoring and why.
Quality threshold
Generated content isn’t valuable just because it’s fast. We aim for work where AI isn’t obvious because of avoidable artefacts or careless decisions.
Experimentation
Experiments are encouraged in controlled conditions, with uncertainty about data, copyright, likeness, safety or bias escalated before a tool is used on live client work.
Review
Reviewed at least every year, and sooner after significant regulatory or platform changes.
Laws and guidance this policy follows
- ICO guidance on AI and data protection
- UK GDPR and Data Protection Act 2018, as amended
- NCSC guidance on secure use of online services and accounts
- Applicable copyright, passing-off, defamation, contract and consumer protection law
The full document
AI & Responsible Technology Policy
Version 1.0 · Effective October 2026 · Owner: Managing Directors / Technical Lead
Questions about this policy?
Questions about how we use AI on your project? Ask us. We’re happy to talk it through.