Answerability state
Authoritative: deep primary evidence, scope and comparison
Defined finding
All eight institutions in the reviewed AI-policy subset publish an explicit boundary or control around candidate or employer AI use in recruitment. Seven of the eight explicitly protect at least one assessment or interview context through an AI restriction or candidate-owned-performance requirement; BCG is not included in that stage-specific numerator because the reviewed guidance does not establish a blanket live-assessment prohibition.
Denominator: 8 · Scope: Defined official-source AI-policy subset reviewed through 13 September 2026 · Do not generalise 8/8, or the stage-specific 7/8 result, to all selective employers. The seven protected-stage policies must not be reported as seven blanket AI bans. Published rules do not establish actual enforcement or hiring effects.
Governed evidence
SIG-108 · McKinsey & Company
McKinsey encourages responsible AI use for resume polishing, interview practice and concept preparation, while prohibiting real-time AI interview answers.
Scope: Global · Interviewing · observed 2026-09-12 · source record SRC-072 Official source
SIG-109 · McKinsey & Company
McKinsey Solve must be completed independently without applications, websites, AI tools or pre-prepared notes unless an approved accommodation applies.
Scope: Global · Solve · observed 2026-09-12 · source record SRC-073 Official source
SIG-111 · UBS
UBS permits AI assistance with application documents where the content genuinely reflects the candidate's own skills, experience and understanding.
Scope: Global · Applications · observed 2026-09-12 · source record SRC-074 Official source
SIG-112 · UBS
UBS requires assessment tasks to be completed independently and prohibits AI assistance.
Scope: Global · Assessments · observed 2026-09-12 · source record SRC-074 Official source
SIG-113 · UBS
UBS prohibits AI-generated responses in recorded or virtual video interviews and prohibits AI recording, transcription or analysis of interviews.
Scope: Global · Interviews · observed 2026-09-12 · source record SRC-074 Official source
SIG-118 · PwC
PwC UK prohibits unauthorised AI or third-party assistance in assessment responses and states it does not use AI to assess Early Careers candidate performance.
Scope: United Kingdom · Early Careers · observed 2026-09-12 · source record SRC-076 Official source
SIG-120 · Clifford Chance
Clifford Chance London assessment days are normally in person and list case or commercial interviews, written work and group exercises for Training Contract and SPARK routes.
Scope: United Kingdom · London Training Contract and SPARK · observed 2026-09-12 · source record SRC-078 Official source
SIG-172 · Deloitte
Deloitte UK permits responsible AI use in preparation, including researching a final-stage assessment topic and practising interviews, while requiring interview responses to be based on the candidate's own experiences and to remain authentic.
Scope: United Kingdom · Early Careers selection process · observed 2026-09-13 · source record SRC-111 Official source
SIG-173 · Deloitte
Deloitte UK states that AI must not be used when completing its online assessments, including the Immersive Online Assessment, OneStop and Job Simulation, which must be completed independently and without AI assistance.
Scope: United Kingdom · Early Careers online assessments · observed 2026-09-13 · source record SRC-111 Official source
SIG-174 · Deloitte
Deloitte UK states that AI must not be used during interviews, whether in person or virtual, and that interviews must be completed without AI assistance.
Scope: United Kingdom · Early Careers interviews · observed 2026-09-13 · source record SRC-111 Official source
SIG-177 · HSBC
HSBC states that candidates may use generative AI to research HSBC and roles, help structure a CV or application where the content accurately reflects their own experience and skills, prepare for interviews, and organise or draft ideas where the final submission remains genuinely their own.
Scope: Global · Recruitment process · observed 2026-09-13 · source record SRC-113 Official source
SIG-178 · HSBC
HSBC states that candidates must not use generative AI to generate responses during live or recorded interviews and must not submit work they cannot personally explain, validate or replicate if asked during recruitment.
Scope: Global · Recruitment interviews · observed 2026-09-13 · source record SRC-113 Official source
SIG-180 · BCG
BCG explicitly permits thoughtful candidate use of generative AI for application support, including brainstorming, structuring a resume or responses, finding sources and practising interview questions, while requiring the candidate's authentic voice, experience and judgement to remain their own.
Scope: Global · Recruitment process · observed 2026-09-13 · source record SRC-115 Official source
SIG-181 · BCG
BCG states that applications are reviewed by experienced recruiters, AI is not used to make hiring decisions, and hiring decisions remain with people; BCG also describes limited employer-side AI support including job recommendations and drafting interview questions.
Scope: Global · Recruitment process · observed 2026-09-13 · source record SRC-116 Official source
SIG-182 · A&O Shearman
A&O Shearman states that candidates may use AI responsibly to complement rather than substitute their skills, including refining CVs, applications and written responses, while requiring honest representation of skills and experience.
Scope: United Kingdom · Early careers recruitment · observed 2026-09-13 · source record SRC-117 Official source
SIG-183 · A&O Shearman
A&O Shearman's London early-careers FAQ states that it does not use AI at any point during its recruitment process and that all assessments and decisions are made by its team.
Scope: London · Early careers recruitment · observed 2026-09-13 · source record SRC-118 Official source
WGI-AI-REVIEW-CC-20260919 · Clifford Chance
The official How we hire page expressly prohibits AI use in the Watson Glaser test. This supports a named-stage restriction, not an inference about every assessment.
Scope: United Kingdom · London early careers: Watson Glaser · observed 2026-09-19 · source record Direct official-source review Official source
WGI-AI-REVIEW-AOS-20260919 · A&O Shearman
The candidate charter asks that applications and assessments reflect the candidate's own thinking, experience and abilities. This is a candidate-owned-performance requirement, not a blanket AI ban.
Scope: Global candidate charter; applied here to UK comparison · Applications and assessments · observed 2026-09-19 · source record Direct official-source review Official source
Limit
Do not generalise 8/8, or the stage-specific 7/8 result, to all selective employers. The seven protected-stage policies must not be reported as seven blanket AI bans. Published rules do not establish actual enforcement or hiring effects.
Evidence route
30-question benchmark · source register · methodology · findings
How to cite
Akram, H., Who Gets In? (Elite Careers Strategy), Where can candidates use AI and where must performance remain candidate-owned?, revised 13 September 2026, https://whogetsin.com/questions/do-elite-employers-ban-ai-in-applications