AI now sits inside a lot of everyday work: screening CVs, scoring video interviews, tracking productivity, summarising meetings. For neurodivergent employees and candidates, that cuts both ways. The same tools that can remove friction for an ADHD or dyslexic employee can also quietly screen out an autistic candidate for not making “enough” eye contact on camera.
This piece looks at where AI tends to go wrong for neurodivergent people, what employers can do about it, and where the technology genuinely helps.
Where AI can work against neurodivergent people
AI systems learn from data. If that data reflects years of hiring and promotion decisions shaped by neurotypical norms, the system will repeat those norms at scale, with an air of objectivity that makes them harder to question. This shows up in three places most often.
Recruitment screening
Many CV screening tools are trained on the profiles of people a company hired in the past. If neurodivergent people were rarely hired, or were marked down for career gaps or unconventional CVs, the tool can learn to filter out similar applicants. Nobody programmed it to discriminate. It simply copied the pattern.
Video interviews and behavioural assessments
Some tools analyse tone of voice, facial expression or word choice to judge a candidate’s “fit”. Autistic candidates may communicate differently, for example with less eye contact, a flatter tone or longer pauses before answering. A model calibrated on neurotypical behaviour can read those differences as low engagement or weak communication, which says nothing about how well the person would do the job. We cover this in more depth in our guide to neurodivergence and AI in hiring.
Productivity and performance monitoring
Monitoring software that tracks keystrokes, active time or response speed assumes everyone works in a steady, linear way. Plenty of people don’t. An employee with ADHD might work in short, intense bursts with breaks in between. A dyslexic colleague might take longer over written messages and be quicker than anyone at spotting the flaw in a plan. Tools that measure process rather than outcome can flag these people as underperforming when they are doing good work. If you are reviewing how you assess people, our neuroinclusive performance management training is a good place to start.
Why regulators are paying attention
This is no longer a theoretical worry. In November 2024 the Information Commissioner’s Office published the results of its audits of companies that develop and supply AI recruitment tools, making almost 300 recommendations. It found that some tools let recruiters filter out candidates with certain protected characteristics, and that some providers estimated candidates’ gender and ethnicity, often from their names, without candidates being told. The full findings are in the ICO’s report on AI tools used in recruitment.
The ICO was looking at data protection, but the lesson for neuroinclusion is the same. Disability is a protected characteristic under the Equality Act 2010, and many neurodivergent people meet the legal definition. If an automated tool puts them at a disadvantage, the employer using it is still responsible, and the duty to make reasonable adjustments applies to recruitment as much as to the job itself.
Five things employers can do
Ask suppliers hard questions about their data
Before buying an AI hiring or performance tool, ask what it was trained on, how the supplier tests for bias, and whether disabled and neurodivergent people were part of that testing. Vague answers are a warning sign.
Rethink what the tool is measuring
Eye contact, speaking speed and “enthusiasm” tell you very little about performance in most roles. Work with suppliers to weight what actually matters for the job, such as problem-solving, accuracy or technical skill, and switch off features that score body language.
Audit regularly
Check who the system is rejecting or flagging, and whether particular groups are over-represented. Once a year is the minimum; a quick check after any change to the tool is better. Our neuroinclusion audits can cover your recruitment process end to end.
Keep a human in the loop
AI should inform decisions, not make them. Recruiters and managers need the confidence and the authority to override an automated rejection, and candidates need a clear way to ask for an alternative format or a human review.
Judge outcomes, not process
If you use monitoring tools at all, point them at what people deliver. Give employees a say in how their work is assessed, and make sure any adjustments you have agreed with them are reflected in the metrics.
Where AI genuinely helps
None of this means neurodivergent people should be wary of AI itself. Many find it useful every day. Speech-to-text and text-to-speech can make reading and writing far less tiring for dyslexic people. AI assistants can break a vague task into steps, draft a first version of a tricky email, summarise a long meeting or turn a page of instructions into a checklist. For a lot of people with ADHD, that removes exactly the kind of friction that used to eat hours.
The difference is control. When the employee chooses the tool and uses it to work their way, AI behaves like any other assistive technology. When the organisation points it at the employee to judge them, the risks above come into play. We look at the positive side in more detail in how AI is becoming a powerful assistive technology for neurodivergent employees.
The bigger picture
AI will keep spreading into hiring and management. Employers who get this right will reach talent their competitors are filtering out without realising it. Systems built around one “normal” way of working will keep missing neurodivergent people; systems designed with difference in mind, and checked by people who understand it, won’t.
How Enna can help
If you are bringing AI into recruitment or performance management, we can look at the process with you and spot where neurodivergent candidates and employees could lose out. Our neuroinclusion maturity assessment gives you a quick benchmark of where you stand today, and our neurodiversity awareness training for managers helps the people making the final call understand what the data might be missing. To talk through your own tools and processes, book a call with our team.
