When the computer says no: AI and ‘meaningful human involvement’ in recruitment

Anyone involved in recruitment knows that the current job market is tough. Many roles are attracting thousands of applications, and the growing use of generative AI tools means that those applications are becoming more homogenous. That can make it time-consuming and difficult to sift out the weaker candidates.

AI is now being used routinely to help employers make the best hiring decisions. Indeed, it’s becoming something of a cliché that nowadays recruitment is just AI-drafted applications being scored by AI tools, with very little human involvement on either side.

There are clear benefits for employers in using AI in recruitment. The technology can review large volumes of applications and CVs quickly and accurately. And despite the very real concerns about bias in AI tools, their use could even aid candidates by removing human bias and error from initial sifts.

Earlier this year, the Information Commissioner’s Office (ICO) published a report into the use of AI and automated decision-making in recruitment. Not surprisingly, the ICO found that technology was increasingly being used to review applications, conduct assessments and in some cases to evaluate responses to interview questions. But recruiters were not always complying with the law in doing so.

Data protection law applies to the processing of personal data, which means information about identifiable individuals. There are specific rules around the use of such data for decisions based solely on automated processing. This covers significant decisions that are made without any meaningful human involvement. A decision is ‘significant’ if it produces a legal or similarly significant effect for the individual. This includes shortlisting and hiring decisions.

The rules on automated decision-making were substantially revised and updated in February this year. The Data (Use and Access) Act 2025 amended the UK General Data Protection Regulation to replace the previous rules with a simpler and more permissive regime. Automated decision-making is now generally permitted, subject to certain exceptions such as where any special categories of personal data are being used, but recruiters must always ensure that certain safeguards are in place. These include being transparent about the use of automated decision-making and giving candidates the opportunity to contest the decision and request that a human reviews the decision.

Employer mistakes

The ICO’s investigation found that many employers mistakenly believed the automated decision-making rules did not apply to their recruitment exercises, because they considered that AI was only being used to support human decisions. However, the ICO found that in reality the human involvement was minimal. This was particularly the case in shortlisting decisions. The ICO’s report emphasises that, unless the human involvement is genuinely meaningful, the automated decision-making rules will apply. For instance, if AI is used to score candidates as part of a shortlisting exercise, there is no meaningful human involvement if the hiring manager merely shortlists those with the highest scores. Instead, the hiring manager must review all scores and exercise real influence over the decision.

Using AI in recruitment

Employers who use or wish to use AI to assist in their recruitment must therefore either ensure there is a genuine and meaningful step where a human makes the decision, or they must follow the rules on automated decision-making. The latter requires them to inform candidates and give them the opportunity to request human intervention. The ICO was generally critical of employers for failing to be fully transparent with candidates about automated decision-making in their recruitment processes.

More fundamentally, the ICO found that employers had not always considered what lawful basis was appropriate for their processing of information about candidates, had not carried out a data protection impact assessment on their use of AI tools and had not fully assessed the fairness of the processing. Where AI tools are being used in a recruitment process, it is important for employers to ensure that those tools are fair. The ICO encourages employers to regularly review their processes for any potential bias.

There is no doubt that AI can have an important role to play in supporting employers in their recruitment activities, and the Data (Use and Access) Act has simplified the rules on automated decision-making. However, the ICO’s report is a timely reminder that employers still need to carefully think through their use of AI tools to ensure they comply with the rules and maintain fairness for candidates.

Data protection training

If you’re looking to improve your data protection knowledge, Blake Morgan’s next BCS practitioner certificate in data protection course will take place in September. We also offer bespoke training on all aspects of data protection compliance.