British companies rushing to integrate artificial intelligence into their products face a growing risk: locking out millions of disabled consumers unless they bring those users into the design process from the start.
That is the central warning from new research by the Business Disability Forum (BDF), based on a poll of 1,032 disabled UK adults conducted with Opinium. Forty percent said co-designing, developing, and testing AI products with disabled people was the single most effective route to genuine accessibility. Close behind came demand for more intuitive interfaces (38%), clearer information about AI’s role in supporting disabled users (37%), and stronger onboarding (36%).
For SMEs weighing how and when to add AI to customer-facing tools, the data carries hard commercial implications. Around one in four UK adults will experience disability at some stage in life — a substantial portion of both the consumer base and the workforce. Building products that fail to serve that audience is increasingly a competitive weakness, not just an ethical one.
The potential upsides are significant. More than a third of disabled adults surveyed said AI tools could improve communications (38%) and online experiences (34%). Additional anticipated gains included better access to healthcare information (33%), education (32%), digital content (32%), support for independent living (31%), customer experience (25%), and employment opportunities (24%).
Yet skepticism persists. Twenty percent said AI products would not help them at all, while a further 18% were uncertain — a meaningful trust gap that companies will need to bridge if adoption is to match the investment pouring into the sector.
A parallel Opinium survey of 2,000 UK adults found broadly similar views across the wider population. Thirty-four percent agreed that co-designing AI with disabled users would boost accessibility — a signal that inclusive design is becoming a mainstream expectation rather than a specialist concern.
Lara Davis, communications director at Business Disability Forum, highlighted the stakes involved.
“There is the potential for AI products and tools to make a radical and positive difference to disabled people’s lives, but there is also the risk that disabled people could be left behind,” she said. “With AI developing at pace and one in four people experiencing disability at some point in their lives, this is not an issue we can afford to overlook.”
Davis called on businesses to actively engage their disabled customers throughout the design, development, and testing phases, alongside offering clearer guidance on the technology generally.
Lucy Ruck, who leads BDF’s Tech Taskforce, urged the same direct approach.
“AI has the capacity to transform lives, but only if we get inclusion right from the start,” she said. “Making sure that disabled people are active participants in shaping this technology isn’t just the right thing to do — it’s how we build AI that genuinely serves everyone.”
BDF has set out four recommendations for businesses and developers. Involve disabled people across the full AI lifecycle — inclusive design benefits all users, not only those with disabilities. Publish clear, accessible information about AI product features in formats suited to different communication needs. Test compatibility with assistive technology rather than assume it. Build ethical oversight and human judgment into both tools and their outputs, drawing on inclusive training data to limit bias and stereotyping.
For founders and product leaders, the advice echoes earlier waves of digital transformation: retrofitting accessibility costs more and delivers less than building it in from day one.