Protecting participants in the Age of AI

21 August 2025

Adam Power-Annand, CEO of Speech Bubbles, shares how he has pivoted his organisation’s approach to how they publicly share images of the work and his journey to putting the protection of children at the heart of publicity materials and policy in today’s context.

From concern to action

Over the last couple of years, I have become increasingly concerned about the myriad of issues that are brought up when using photographs of our participants on websites and social media - including issues around copyright, ownership, and informed consent; especially when it came to children and young people. This was before the rise of generative A.I. and the speed and ease with which any image can be captured, manipulated and repurposed. When my third grandchild was born; my daughter requested that we not share images of her beyond a select few family members, and I realised that we had to move from concern to action.

The nature of work

Speech Bubbles is a drama intervention for young children who have been referred, which supports communication, confidence and social wellbeing. It’s a joyful practice with every session full of movement and sound as small groups of children become a company of storytellers and performers, creating and acting out each other’s stories. With no props, costumes or instruments, these companies act out stories set in forests, oceans, space or even their own front rooms. Each story is filled with amazing characters such as grandmas, the police, a ravenous shark, Spider-Man and mesmerising events; chases around the world, playing in the swing park or eating the biggest ice cream ever.

Because it happens in a small group setting, all these weekly performances disappear as soon as they are told.

“I regard the theatre as the greatest of all art forms, the most immediate way in which a human being can share with another the sense of what it is to be a human being.” Oscar Wilde.

Communicating our work to the world

At Speech Bubbles, we want to capture the humanity, joy, movement, the acting and share it more widely. In the age of photography, websites and social media we’ve done that by taking pictures and adding captions. ‘Look at these children having fun’, ‘look at this drama person facilitating that fun’, ‘here we are in a classroom where theatre is happening!’ The best pictures capture the children in the moment of playing a character. We are drawn to their facial expressions as they act out the anger, confusion, laughter or happiness that the character is feeling.

We have always taken care to make sure images celebrate the children and showcase their creative endeavours. We ask for permission from the children and their parent/carers to use the image and after a couple of years we accept that consent has run out and then collect new images.

With the rise of generative A.I. we are looking at those images afresh and changing tack. We have been careful to curate them, to celebrate the children and the work. We have also been careful to gain consent for their use, but now when we look at those images we also see something different. Something less celebratory, maybe less human; we see data that can be stolen and manipulated and can be quickly and easily subverted or twisted. Until legislation and technology catch up, we are no longer comfortable showing images of children’s faces, and we no longer believe that we can ask for consent for images to be used in a limited way.

Doing it differently

I wanted to move quickly on this, and with the wider support of my General Manager and the board of trustees, set to work. We decided that we no longer wanted identifiable photographic images on our website or social media.

Steps 1-3 below had minimal cost beyond staff time, step 4 will have a direct associated cost.

Step 1: Immediate response

Speech Bubbles was being represented by a young person, Daniel, at the House of Lords for the launch of the Speech, Language and Communication Alliance. To celebrate and share this, we commissioned an artist to do an illustration of Daniel speaking and shared that alongside a transcription of his speech. The artist worked from a simple verbal prompt. Learn more.

Step 2: Audit

We wanted to know what images we were using and where. Which of these images were identifiable, and which showed the children’s faces? This revealed a couple of issues. Firstly, images that were directly on the website that could easily be removed, and secondly, images embedded in other documents.

Step 3: Replacement images

We trawled through our photographs and other artistic representations of the project, and with some careful cropping were able to put together a collection of usable images. These were used to do a quick refresh of our website.

We are now working through historic documents, blog posts and reports and changing or removing photographic images.

Step 4: Commission new artworks

In the Autumn, we will put out a call for an artist to create a new series of Speech Bubbles images. These will be used across our website, resources, reports and across social media. The contract will ask the artist to confirm that they haven’t used AI in the creation of the works.

We have filmed content of Speech Bubbles that includes ‘live action’, and we will commission an animator to either create a new film or to animate that specific section.

Step 5: Keep learning

By reaching out to our networks and by sharing our concerns and actions, we are raising the issue and making immediate changes. We are not experts in the technical and legislative side of this, so we need to stay alert and connect with the people who are more knowledgeable than us.

Things you can do

1. Get informed, read this blog from NSPCC
2. Have conversations: with your team, practitioners, board and any partners you work with about A.I.
3. Rework policy: We have an annual review of our Safeguarding and image policies; these concerns will be written into our next update.
4. Take all identifiable and facial images of children & young people off your website and promotional material.
5. Join the conversation; [add link to roundtable discussion/working group]

Connect with Adam Annand, CEO of Speech Bubbles via Instagram, Bluesky or LinkedIn