You’ve polished your CV, spent two evenings writing a cover letter that actually sounds like you, and hit submit. Then comes the email. Before we invite you to interview, we’d like you to complete a short video assessment. Fifteen questions. No interviewer. Just you, a webcam, and software that will analyse your facial expressions, word choice, speech patterns, and tone of voice. Welcome to the new world of AI job screening that UK employers are quietly rolling out across almost every sector.
I’ve spoken to candidates who had no idea this was happening to them until after the fact. One graduate told me she’d completed a ten-minute video interview for a logistics company in Birmingham and only found out afterwards that the platform she’d used, HireVue, had scored her on traits including “enthusiasm” and “conscientiousness” before a single human had seen her face. She didn’t get the job. She has no idea why.
What are these AI screening tools actually doing?
The platforms vary, but the broad approach is consistent. Tools like HireVue, Pymetrics, Retorio, and the UK-grown Sapia.ai ask candidates to answer questions on video or via text chat. The AI then scores them across a range of personality dimensions, typically drawing on psychometric frameworks like the Big Five personality traits (openness, conscientiousness, extraversion, agreeableness, and neuroticism) or proprietary models the companies have built themselves.
Some go further. Certain platforms claim to measure “cognitive ability” through games or pattern-recognition tasks. Others do sentiment analysis on written answers. A few still use facial expression analysis, though this has become legally and reputationally risky enough that several providers have walked it back publicly.
The pitch to employers is compelling: faster shortlisting, lower unconscious bias from human recruiters, and a consistent candidate experience at scale. For a company receiving 4,000 applications for 20 graduate roles, the logic is obvious. A human cannot meaningfully review that volume. A machine can, and in theory it does so without getting tired, grumpy, or distracted by a candidate’s name or accent.
In theory. The practice is messier.
Does AI job screening actually work?
This is where things get genuinely complicated. The scientific basis for some of these tools is shaky. The BBC has previously reported on research suggesting that facial expression analysis, in particular, rests on assumptions about emotion and behaviour that psychologists widely dispute. If your cultural background, neurological differences, or simply your nerves cause you to present emotions differently, a system trained predominantly on one demographic could score you poorly through no fault of your own.
Pymetrics, to its credit, publishes bias audits. Others are less transparent. When a tool is described as proprietary, that usually means candidates and employers alike have no real way to interrogate what it’s actually measuring. The Information Commissioner’s Office (ICO) has published guidance making clear that automated decision-making in recruitment must comply with UK GDPR, including the right not to be subject to decisions based solely on automated processing without meaningful human review.
Your legal rights as a job applicant
This is the bit most candidates don’t know about, and honestly, it’s quite encouraging. Under UK GDPR (retained in domestic law post-Brexit as the UK GDPR), you have real rights here. If a decision about you is made solely by automated means and produces a significant effect, like being rejected from a job, you are entitled to request human review. You can also ask the employer to explain the logic behind the automated decision and object to the processing.
The ICO’s guidance is explicit: employers using AI job screening tools cannot simply hide behind “it’s our system” as a justification. They need to be able to explain what the tool measures, how decisions are weighted, and what safeguards exist against discriminatory outcomes.
Under the Equality Act 2010, if a screening tool produces outcomes that disproportionately disadvantage candidates of a particular race, gender, disability status, or age (among other protected characteristics), that could constitute indirect discrimination. The burden of proof shifts to the employer to justify the tool as a proportionate means of achieving a legitimate aim.
In practice, most candidates don’t know to ask. And employers are banking on that.
How to handle an AI screening request
My honest take: don’t panic, but don’t be naive either. A few things worth knowing before you hit record.
First, you can ask. Before completing any video assessment, email the recruiter and ask whether the platform uses automated scoring, what it measures, and whether a human will also review your responses. You’re entitled to this information under UK GDPR. Some employers will be refreshingly upfront. Others will waffle. The response itself tells you something about the company culture.
Second, you can object. If you believe you’ve been unfairly rejected through an automated process, you can write to the employer invoking your Article 22 rights under UK GDPR, requesting that a human reviews your application. They’re legally obliged to respond. Whether they’ll then offer you the job is a different matter, but you do have that lever.
Third, if you have a disability that might affect your performance on video or in a timed cognitive assessment, you can request reasonable adjustments before completing it. The same employer obligations that apply in a face-to-face interview apply here.
What this means for the future of hiring
AI job screening isn’t going away. Recruitment is expensive and slow, and these tools genuinely do solve real logistical problems for large employers. The NHS, major UK banks, the civil service, and dozens of graduate schemes now use some form of automated pre-screening. That’s not a niche trend; it’s mainstream.
What’s lagging is accountability. Right now, there’s no UK-wide mandatory audit requirement for recruitment AI, no register of which employers use which tools, and no independent body that a rejected candidate can easily complain to. The ICO has enforcement powers but limited resource to pursue individual recruitment cases.
Some of this parallels the broader pattern we’ve seen with regulators scrambling to keep pace with platforms that move faster than the rules. And just as UK councils are struggling to cope with AI-generated planning applications, HR departments are often equally unprepared to understand the tools they’re deploying, let alone explain them to rejected candidates.
The best employers are starting to treat these tools as a first filter, not a final verdict, always pairing AI scores with human review. The worst are outsourcing decisions entirely to a system nobody inside the business fully understands.
If you’re going through a job search right now and land one of these assessments, you’re not powerless. You have rights, you have recourse, and you’re allowed to ask questions. Ask them. The employers worth working for will answer.
Frequently Asked Questions
Is AI job screening legal in the UK?
Yes, but with conditions. UK GDPR requires employers to tell you when automated decision-making is being used and to provide a way for you to request human review of any significant decision. Simply relying on an AI score without meaningful human oversight is not compliant.
Can I refuse to complete an AI video interview?
You can, but the employer may simply move on to the next candidate. A better approach is to ask questions about the tool before completing it, or to invoke your right to request adjustments if you have a disability that affects your video performance.
What does HireVue actually measure?
HireVue analyses verbal content (your word choices and answers), voice characteristics (pace, tone, energy), and in some configurations, facial expression data. It maps these against a behavioural model built from previous successful hires at that employer, though the exact weighting is proprietary.
Can I find out why I was rejected by an AI screening tool?
Under UK GDPR Article 22, you can ask the employer to explain the logic behind an automated decision and request that a human reviews it. You’ll need to write to them directly, citing your data subject rights, and they are legally obliged to respond within one month.