Over three quarters of the UK public want child safety checks on new generative AI products

Safety concerns around generative AI

New research commissioned by us highlights the different ways that generative artificial intelligence (AI) is being used to groom, harass and manipulate children and young people.

This comes as polling shows that the UK public are concerned about the rollout of AI. Savanta surveyed 3,356 people from across the UK and found that most of the public (89%) have some level of concern that “this type of technology may be unsafe for children”. ¹

The majority of the public (78%) said they would prefer to have safety checks on new generative AI products, even if this caused delays in releasing products over a speedy roll-out without safety checks.

Our new paper shares key findings from research conducted by AWO, a legal and technology consultancy. The research Viewing Generative AI and children’s safety in the round² identifies seven key safety risks associated with generative AI; sexual grooming, sexual harassment, bullying, financially motivated extortion, child sexual abuse & exploitation material, harmful content, and harmful ads and recommendations.

Generative AI is currently being used to generate sexual abuse images of children, enable perpetrators to more effectively commit sexual extortion, groom children and provide misinformation or harmful advice to young people.

From as early as 2019, we have been receiving contacts from children via Childline about AI.

Solutions and urgent actions

Our paper outlines a range of different solutions to address these concerns including stripping out child sexual abuse material from AI training data and doing robust risk assessments on models to ensure they are safe before they are rolled out.

The government is currently considering new legislation to help regulate AI and there will be a global summit in Paris this February where policy makers, tech companies and third sector organisations, including the NSPCC and their Voice of Online Youth, will come together to discuss the benefits and risks of using AI.⁴

We’re calling on the government to adopt specific safeguards for children in its legislation. Four urgent actions are needed by the government to ensure generative AI is safe for children:

1. Adopt a duty of care for children’s safety
   Generative AI companies must prioritise the safety, protection, and rights of children in the design and development of their products and services.
   
   
2. Embed a duty of care in legislation
   It is imperative that the government enacts legislation that places a statutory duty of care on generative AI companies, ensuring that they are held accountable for the safety of children.
   
   
3. Place children at the heart of generative AI decisions
   The needs and experiences of children and young people must be central to the design, development, and deployment of generative AI technologies.
   
   
4. Develop the research and evidence base on generative AI and child safety
   The government, academia, and relevant regulatory bodies should invest in building capacity to study these risks and support the development of evidence-based policies.