An AI hiring agent at law firm Addleshaw Goddard ranked unusually polished CVs highly, while a working paper suggests candidates using the same AI model as a recruiter may be more likely to make the shortlist. The findings raise a practical concern for applicants and employers: AI-written applications could gain an advantage for reasons that have little to do with the candidate’s ability.
Watch Desk analysis
What happened
The Irish Times account describes an experiment in which Addleshaw Goddard’s HR team asked an internal Microsoft Copilot agent to assess applications for an IT job. There were 178 applications. The agent and the human reviewers produced different shortlists; the agent favoured some CVs making impressive claims despite limited experience. The successful candidate was chosen by people, not the agent.
The account also describes a working paper simulating hiring processes for 24 roles. Its researchers found that applicants using the same large language model as the recruiter were between 23 and 60 per cent more likely to be shortlisted than applicants whose CVs were written without AI. The paper is still being revised for peer-reviewed publication, so those results may change.
Why it matters
If a hiring tool rewards signs of AI use, applicants may feel pushed to use AI simply to avoid being disadvantaged. That risks turning recruitment into an arms race in which polished applications travel further, while employers learn less about the people behind them.
The reported study is not proof that every hiring system prefers AI-written CVs. But it gives employers a reason to test for the effect rather than assume automated screening is neutral. The paper suggests two possible ways to reduce the bias: instruct models to assess CV content without considering its origin, and use several assessment models rather than relying on one.
Our read
The Addleshaw Goddard example is a useful warning, not a verdict on AI hiring. Its agent’s shortlist differed from the human reviewers’, and the people made the final choice. That is precisely why employers should check what their systems reward before letting a tidy score quietly become the decision.
For jobseekers, the uncomfortable prospect is that avoiding AI may itself become a disadvantage. Employers using these tools should be able to explain how they test for that, and what human review can change.
What to watch
- Whether the working paper’s findings hold up as it is revised and reviewed.
- Whether hiring-tool providers test for preference towards CVs written with their own models.
- Whether employers disclose how automated scores affect shortlisting and final decisions.
Discussion spark: Should employers ban AI-written CVs, or should they instead be required to prove that their screening tools do not favour them?
Sources and evidence
- Source update (12 October 2026, 05:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.