Somewhere between submitting your CV and receiving that politely worded rejection email, a decision was probably made about you. Not by a recruiter, not by a hiring manager, not by anyone you could ever sit across a table from. An algorithm looked at your application, scored it against a set of criteria you were never shown, and quietly moved you to a different pile. This is the reality of AI recruitment bias in the UK, and it is happening at a scale that most jobseekers simply do not know about.
According to a 2025 report from the Chartered Institute of Personnel and Development, more than 40% of large UK employers now use some form of automated screening at the initial CV stage. The number has grown sharply since 2022, driven by cost pressures, the volume of applications for popular roles, and a genuine belief among HR departments that algorithms are more consistent than tired humans sorting through hundreds of near-identical documents. Whether that belief is justified is a rather different question.
What these tools actually do
The terminology can be confusing. “AI recruitment tools” covers a wide range of technologies: applicant tracking systems that parse CVs for keywords, machine learning models trained on historical hiring data to predict candidate success, video interview software that analyses facial expressions and speech patterns, and chatbot pre-screening tools that ask candidates a series of questions before a human is ever involved. Some employers use one of these; some use all four in sequence.
The promise is efficiency. A major retailer receiving 15,000 applications for a graduate scheme simply cannot have a recruiter read each one. Automated systems can filter that to a manageable longlist in seconds. The problem is that every tool trained on historical data inherits the biases embedded in that data. If a company’s most successful hires over the past decade have disproportionately come from certain universities, lived in certain postcodes, or used certain vocabulary in their covering letters, the model learns to weight those signals positively. Without meaning to, the system encodes and then amplifies existing inequalities.
The documented biases already on record
This is not theoretical. The Equal Opportunities Commission, now subsumed into the Equality and Human Rights Commission (EHRC), documented cases in the early 2020s where automated shortlisting tools were screening out candidates with career gaps, a pattern that disproportionately affected women who had taken time away from work to raise children or care for elderly relatives. The EHRC’s 2024 guidance on algorithmic decision-making in employment noted that bias in AI hiring tools can constitute unlawful indirect discrimination under the Equality Act 2010, even when the tool itself makes no reference to any protected characteristic.
Video interview analysis software has attracted particularly sharp scrutiny. Several products sold to UK employers claim to assess a candidate’s suitability by analysing vocal tone, eye contact, and facial micro-expressions. There is no robust scientific evidence that these signals reliably predict job performance. What is documented is that they perform differently across ethnic groups, disadvantage candidates with certain disabilities or neurodivergent conditions, and penalise anyone whose first language is not English. The RNIB and Scope have both raised formal concerns about specific platforms. This sits alongside the broader workforce crisis that I wrote about when covering how automation is restructuring entire layers of the UK workforce, removing human judgement from decisions that once required it.
What the ICO and EHRC are actually doing
The Information Commissioner’s Office entered this space via data protection law rather than employment law. Under UK GDPR, candidates have the right not to be subject to a decision based solely on automated processing that produces a legal or similarly significant effect. Recruitment is explicitly cited as an example of such an effect. The ICO’s guidance, updated in 2025, makes clear that employers using fully automated screening must inform candidates this is happening, offer them the right to request human review, and be able to explain how the system works in terms the candidate can understand.
In practice, very few employers are meeting all three requirements. The ICO issued enforcement notices to two unnamed large employers in late 2025, the first such action specifically targeting recruitment automation. The EHRC has separately signalled that it plans to pursue test cases under the Equality Act where there is evidence of systematic algorithmic discrimination. The ICO’s full guidance on AI and data protection is publicly available and worth reading in detail if you work in HR or are building these tools.
The honest assessment is that regulatory action is moving slowly relative to the pace of adoption. The tools are being deployed. The legal framework exists. The enforcement machinery is grinding into motion. The gap between those three facts is where most of the harm is currently occurring.
How the technology connects to wider workforce pressures
It is worth situating this within a broader context. Britain’s labour market is under real structural pressure. The cost-of-living squeeze has pushed more people into job searching at the same time as employers have cut recruitment budgets. Middle management has been thinned out, as I explored when looking at the hollowing out of British management structures, which means fewer experienced people are available to run traditional recruitment processes. The result is that automation fills a gap left by deliberate cost-cutting, and then its failures are presented as a technology problem rather than a resourcing decision.
There is also a growing industry around helping candidates beat these systems. CV optimisation services promise to keyword-stuff your application so it scores well with automated parsers. Some of this advice is genuinely useful. Knowing that your CV will be read by software before it is read by a person is information worth having. Tailoring your language to match the exact phrasing in a job advert is simply sensible. But there is something genuinely uncomfortable about a hiring ecosystem that rewards people who understand how to game an algorithm over those who are simply qualified and honest about their experience.
Practical guidance for jobseekers in 2026
If you are currently applying for roles in the UK, a few things are worth knowing. First, you can ask any employer whether automated tools were used in your assessment. Under UK GDPR, if automated processing was involved in a significant decision about you, you have the right to know. Second, if you reach the interview stage and are asked to complete a video interview analysed by software rather than a human, you can ask for an alternative format as a reasonable adjustment, particularly if you have a disability or a condition that might affect how the software interprets your presentation.
Third, pay close attention to job adverts and mirror their language in your CV. Not because you are gaming anything, but because if a recruiter wrote “stakeholder engagement” rather than “client management”, their ATS has probably been trained on that term. Fourth, keep a record of every application. If you are consistently screened out at the first stage despite strong credentials, that pattern could be relevant if a systemic bias complaint ever reaches the EHRC or a tribunal.
I would also suggest being sceptical of the CV services that charge significant fees to “ATS-proof” your application. The genuinely useful advice in this space is free or near-free. What these services often sell is anxiety, packaged as expertise. Some sectors, incidentally, have started to push back against over-automation. Engineering and manufacturing firms, in particular, have found that rigid keyword filters screen out talented people who came to their skills through unconventional routes. Rapid manufacturing services like Print Shape represent the kind of specialist technical environment where a human recruiter’s ability to read between the lines of a CV still matters enormously.
Where this is heading
The EU AI Act, which came into force in 2024, classifies AI systems used in employment decisions as high-risk, requiring transparency, human oversight, and regular bias auditing. The UK government has not adopted the EU AI Act post-Brexit, and its own AI governance framework, published under the previous government, relies heavily on existing regulators rather than new legislation. The ICO and EHRC have the powers they need. The question is whether they have the resources and the political will to use them consistently.
My reading of the current position is that the next two years will see either a significant enforcement moment, one high-profile case that forces the industry to take this seriously, or a gradual normalisation of the status quo in which AI hiring bias becomes an accepted feature of the landscape rather than a scandal. Which of those happens depends partly on jobseekers knowing their rights and exercising them, and partly on whether the EHRC chooses to pursue the test cases it has signalled. Watch both carefully.
Frequently Asked Questions
Is it legal for UK employers to use AI to screen CVs without telling candidates?
Under UK GDPR, employers must inform candidates if a solely automated process was used to make a significant decision about them, such as whether to progress their application. They must also offer the right to request human review. Many employers are not currently compliant with these requirements, and the ICO issued its first enforcement notices in this area in late 2025.
What is AI recruitment bias and why does it happen?
AI recruitment bias occurs when automated screening tools produce systematically unfair outcomes for particular groups of candidates. It typically happens because the models are trained on historical hiring data that reflects past human biases, causing the algorithm to replicate and entrench patterns of discrimination even without any explicit instruction to do so.
Can I request a human review of my rejected job application?
If a UK employer used solely automated processing to reject your application, you have the right under UK GDPR to request that a human reviews the decision. You should make this request in writing to the employer’s HR or data protection contact. They are legally obliged to respond.