For journalists · live evidence deskLatest selection evidence
Current reporting leads drawn from official institutional sources. Each note is deliberately narrow: it says what the source supports, identifies what it does not establish, and gives reporters the primary pages to check.
Updated 12 September 2026 · research lead Hassan Akram
Banking · AI fluency and assessment boundaries
UBS pairs graduate AI fluency with explicit no-AI boundaries in assessments and interviews
UBS's current Graduate Talent Program includes an AI Fluency Pathway focused on real-world use, responsible application and sound judgement. Its current application guidance allows AI support for application documents but prohibits AI assistance in assessments and interviews.
Reporting boundary: this is an institution-specific example. It is not evidence that the banking market as a whole has adopted the same model.
Primary sources: UBS Graduate Talent Program · UBS AI usage guidance
Corporate law · AI and authorship
Two current law-firm policies permit AI support while protecting authentic candidate judgement
Covington's London guidance permits AI support while requiring final submissions to reflect the candidate's own original responses and voice. Broadfield permits AI for research, structure or proofreading but warns against over-reliance because it assesses genuine personal experiences and authentic reflections.
Reporting boundary: these are two institution-specific examples consistent with the WGI ownership/authenticity thesis. They do not establish prevalence across legal recruitment.
Primary sources: Covington London applications · Broadfield early careers
Corporate law · selection architecture
Clifford Chance uses AI inside an immersive route that can fast-track candidates to a Training Contract Assessment Day
Clifford Chance's The Challenge asks candidates to collaborate, use AI tools and tackle trainee-style legal and business problems for a fictional global client. Standout performers in the final assessed task can be fast-tracked to a Training Contract Assessment Day.
Reporting boundary: this is a current Clifford Chance pathway, not evidence that law-firm selection generally has moved to AI-enabled simulation.
Primary sources: Clifford Chance The Challenge · London Training Contract
Methodological ruleWho Gets In? distinguishes observed institutional evidence from market prevalence. A small number of examples remains a small number of examples until the governed register supports a broader denominator.
