If candidates are using AI, is it time to rethink recruitment?

Artificial Intelligence is rapidly reshaping the job market. Research we conducted last year found that 54% of charity sector job seekers already use AI in their current roles, and 60% have used it when applying for jobs, yet only 11% of organisations use AI in recruitment processes. That gap is creating real challenges for hiring managers.

A regular question we are posed is “How do we spot AI?” , however perhaps a better question is “If traditional application methods can be undermined so easily, were they ever the best measure of capability?”

The problem with traditional assessment methods

Cover letters and supporting statements were designed for a world where candidates wrote everything themselves. Today, AI tools can generate job‑tailored statements in minutes – and not in a generic style, but increasingly in the tone and voice of the applicant. That makes it much harder to judge authenticity or assume strong communication simply from polished words on a page.

Our research highlights this shift:

• 83% of candidates using AI do so to enhance cover letters or supporting statements
• 37% admit writing those statements entirely with AI.

That makes it increasingly hard to infer judgment, writing ability or capability from the quality of an application alone - especially in a sector where supporting statements still carry weight in shortlisting. And it exposes a truth that’s always been there: candidates skilled at writing have long had an advantage over those equally strong in practice but less articulate on paper.

Viewed through that lens, AI might not be the problem. It might simply be removing the illusion that supporting statements were ever a perfect proxy for performance.

AI detection: a false comfort

A natural reaction is to reach for AI-detection tools. But the reality?

• 60% of organisations are not even asking applicants to disclose AI use.
• Where detection tools are used, results can be unreliable; widely used checkers have produced inconsistent outcomes - flagging genuine writing as AI while missing AI-generated content.

That makes using an AI-detection score as part of a recruitment decision problematic. Employers risk discounting genuine applications based on a technology that cannot reliably establish how a piece of writing was produced.

There are also important fairness and data protection considerations. If technology is being used to analyse candidate information or influence recruitment decisions, employers need to understand how it works, how accurate it is and whether it could disadvantage particular groups. Candidates should also be given appropriate transparency about how their information is being used.

This is particularly important where candidates may use AI tools for legitimate reasons - for example, to help structure their thoughts, improve written English or support accessibility needs.

Rather than trying to police AI use through imperfect detection tools, a more robust approach is to design recruitment processes where candidates have opportunities to demonstrate and substantiate their own experience, judgement and capability.

Even if detection technology improves, employers still need to ask if spotting AI is really where energy and budget should go or if the bigger challenge is how we assess capability in a world where AI is normal?

Why banning AI may backfire

Some employers are turning to AI bans. But enforcing them consistently can be extremely difficult.

Paradoxically, AI literacy is now part of the skills many roles require. If an employee will reasonably use AI to draft content, research or streamline admin, why design assessments that exclude it?

That doesn’t mean unrestricted AI use in every stage. If you specifically need to test unaided writing or technical capability, that’s valid. But the principle is simple:

Be clear on what you’re assessing and why.

Our research underscores why rules matter:

• 40% of organisations currently have no AI guidelines
• 53% of those already using AI do so without any restrictions on platforms.

That raises important questions around data protection, organisational compliance and fairness.

What should organisations assess instead?

Rather than “did this candidate use AI?”, ask:

• Can they explain their reasoning?
• Can they apply experience to new challenges?
• Do they show sound judgment and alignment with your organisational values?
• Have they spotted and corrected AI errors?

Don't forget that increasingly, being able to use AI responsibly may be a skill worth testing.

Practical alternatives to cover letters

If traditional applications don’t tell you enough, build in steps that reveal real-world capability:

• Work samples or simulations that mirror actual tasks.
• Structured interviews that explore decisions, collaboration and credibility of evidence.
• Scenario-based discussions testing integrity and judgment.
• Portfolio reviews to assess tangible outcomes.
• Guided reflection: ask candidates what tools they used, how and why.

Timed or supervised tasks can still serve a purpose where authenticity is essential, provided reasonable adjustments keep tests fair and relevant.

Balancing rigour with reality

AI is contributing to higher application numbers across many areas, but so are wider economic pressures. That leaves employers juggling fairness with feasibility. Adding multiple stages may feel like the answer, but more steps don’t always mean better hiring. Instead:

• Front-load clarity – make expectations explicit before candidates apply
• Use lightweight, high-signal tools early – for example a brief scenario question rather than a full exercise
• Automate where appropriate (e.g. scheduling) so human time is protected for decisions that matter most

In short: prioritise quality over quantity.

Be clear with candidates

Without doubt, transparency is the simplest improvement most organisations can make. Avoid vague “AI is not allowed” rules and instead:

• Define where AI can assist (e.g. structuring an application)
• Define where it cannot (e.g. tasks designed to assess unaided reasoning), and explain why (e.g. to ensure fairness where writing ability is an essential requirement or to test problem-solving without tools).
• State a clear principle: applications must be authentic - real experiences, true examples, and the candidate’s own voice

Some organisations phrase it like this:

“AI tools may support how you express your experience, but the ideas and examples must come from you. We want to understand the real you, not a model answer.”

Candidates already expect to justify their applications - so signalling the standard upfront helps protect fairness for all.

The bottom line

The challenge isn’t spotting AI. It’s designing recruitment processes that still identify judgment, adaptability and values as AI use becomes normal.

For charities, that likely means giving less weight to supporting statements - and more to behaviours, evidence and applied skills.

AI hasn’t broken recruitment; it has exposed weaknesses that were already there. Rather than redesigning processes to “catch” AI, let’s shape them so AI doesn’t hide the best candidate.

Three actions for employers now

1. Review what you’re really measuring

If supporting statements no longer give you meaningful insight, reduce their weight. Bring in at least one assessment stage that shows applied judgment or problem-solving.

2. Publish clear AI guidance

State where AI is acceptable, where it isn’t, and why. Clarity protects fairness, strengthens trust, and supports fairness, consistency and trust.

3. Focus on quality over quantity

With more applications (thanks to AI and the economy), resist adding complexity for complexity’s sake. Use short, high-signal tasks early and save human time for decisions that really matter.