REPORT
The state of AI in talent assessments
A playbook for ethical AI transformation in assessment, hiring and development
Contents:
- The perception gap in AI assessment
- The cheating arms game
- Using tools we don’t fully understand
- Unfairness and the audit gap
- What HR leaders actually want
- From fast hiring to fair hiring
- The safe AI checklist
Download the report
Talent assessment has entered a new phase. Adoption is widespread, AI is already embedded, and yet confidence has not kept pace. The result is a clear perception gap between how these tools are being used and how much they are trusted.
The data reflects this tension. 94% of organisations surveyed use talent assessments and over 50% confirm AI is involved, understanding remains uneven. A significant proportion operate in a ‘shadow AI’ zone, where tools are in use without full visibility or control. At the same time, concern is rising. Many believe candidates are using AI to manipulate results, and a growing number have already seen evidence of this in practice. Concerns around unfair or inaccurate evaluations sit close behind.
This is not a new behavioural problem, but it is a new operational reality. The motivation to perform well in assessments has always existed. AI has changed the scale, speed, and accessibility of that behaviour. In parallel, organisations are under pressure to adopt AI to improve efficiency and decision-making. This creates a dual acceleration. Candidates are adapting quickly, and so are employers, often without a shared understanding of the risks.
The opportunity remains significant. AI has the potential to make assessment faster, more consistent, and more robust than manual approaches, while freeing up time for human-led evaluation where it matters most. However, the data shows that confidence in ethical use is still developing. Only a small proportion of organisations report being highly confident in how AI is applied within their assessment processes.
This gap between adoption and understanding is where risk accumulates. Concerns around cheating, fairness, legal exposure, and transparency are closely linked. Addressing them requires a structured approach. Scientific validity must remain the foundation. AI should be used to enhance, not replace, decision-making. Human oversight must ensure accountability at every stage.
This report explores how organisations are navigating this shift, where confidence is growing, and where greater control is needed to maintain trust and integrity in AI-driven assessment.
Among 382 HR and talent professionals surveyed, 94 % use talent assessments
51% of those respondents confirmed that AI plays a role in how those tools are delivered or scored.
But alongside adoption sits a persistent unease. Only 22% describe themselves as ‘very confident ’ that AI is being used ethically.
Part 1: Reluctant reliance - using tools we don’t fully understand:
Talent assessments are now widely used, but understanding of AI within them is uneven. Around one third of organisations operate in a ‘shadow AI’ state, using systems they cannot fully explain or audit.
Adoption is being driven faster than governance, increasing risks around accountability, bias, and legal exposure. As a result, many organisations are relying on tools that are shaping decisions without clear visibility or control.
are operating with what we call ‘Shadow AI’: algorithms influencing talent decisions without full knowledge or consent.
The rise of Shadow AI
‘Shadow AI’ describes the use of AI in talent assessment without full visibility or understanding by the organisations deploying it.
For candidates, this means decisions about hiring, promotion, or development may be influenced by opaque algorithms, making it difficult to challenge outcomes or understand how their performance is evaluated.
For organisations, it introduces significant risks: accountability is blurred, bias detection becomes challenging, and legal or reputational exposure rises if outcomes are questioned.
More broadly, the assessments industry faces pressure to balance rapid adoption with transparency, as widespread Shadow AI undermines trust in both the tools themselves and the decisions they produce, highlighting the urgent need for governance, auditability, and ethical standards.
Shadow AI is already making decisions
A significant portion of hiring is influenced by AI that organisations cannot fully see or explain. This creates risk across accountability, bias, and compliance.
Part 2: The double threat - unfairness and the audit gap
Limited visibility into how AI is used in assessments creates a dual risk that organisations must manage. One sits within the tools themselves, where lack of transparency can lead to unclear decision-making, hidden bias, and reduced accountability. The other comes from candidates, who are increasingly using AI to influence or manipulate outcomes. Together, these forces create a more complex risk environment where both the system and its users can undermine the integrity of assessment if not properly controlled.
have already seen evidence of candidate cheating. A further 36% believe it is likely happening in their organisation.
The cheating arms race
AI is accelerating a hiring arms race, where organisations and candidates are both adapting faster than the systems designed to control them.
For candidates, AI is increasingly used to enhance or shape assessment performance, especially in written formats. However, not all assessments are equally exposed. Tools that measure behaviour or real capability remain harder to game, forcing a balance between optimisation and authenticity.
The cheating arms race has started
Candidates are using AI to optimise outcomes, forcing organisations to rethink how assessments are built, secured, and interpreted.
Part 3: The trust protocol: what HR leaders really want
Trust is now central to AI-driven assessment. HR leaders still value validity and efficiency, but expect transparency, evidence, and strong candidate experience alongside them. Confidence is no longer built on claims alone. It requires clear explanation, validation, and assessment design that is fair and harder to manipulate. AI is being accepted as a support tool, reinforcing the need for visible human oversight.
want clearer explanations of how AI is used in the designing delivery of their assessments.
What HR wants
Trust in AI assessment depends on proof, transparency, and design.
What it means for candidates
Candidates want clarity on how they are assessed and expect processes that feel fair and understandable. Well-designed assessments reduce the incentive to cheat and increase trust in outcomes.
What it means for organisations
Organisations must move beyond vendor claims and show how their tools work, backed by evidence. Transparency is now expected, with demand for explainability, auditability, and compliance. Candidate experience acts as a control mechanism, reducing manipulation while improving engagement. Assessment design should shift towards more interactive formats. AI should remain a support tool, with clear human accountability.
From fast hiring to fair hiring
Speed is no longer the priority. The focus is shifting toward skills, fairness, and processes that genuinely predict performance.
Part 4: The road ahead
Talent assessment is shifting from a focus on speed to a focus on quality, fairness, and real capability. As AI reduces the value of traditional signals like CVs, organisations are moving toward skills-based evaluation that is harder to manipulate and more predictive of performance. This marks a transition from rapid adoption to more deliberate, outcome-driven hiring.
Shifting to skills based hiring
Hiring is moving from fast to fair, and from CVs to capability.
What it means for candidates
Candidates are increasingly assessed on what they can actually do, rather than how well they present themselves on paper. This creates a fairer environment, but reduces the ability to rely on polished applications or AI-assisted responses alone.
What it means for organisations
Organisations must rethink how they identify talent, shifting away from CV-led screening toward direct measurement of skills and behaviour. Fairness is becoming commercially important, influencing both candidate trust and offer acceptance rates. Transparent, evidence-based processes will attract stronger candidates and improve conversion. Assessment tools need to focus on real capability, reducing reliance on signals that AI can easily enhance or distort.
