Research programme of Elite Careers StrategyUse the finding. Check the evidence. Keep the limit.
Who Gets In? is the official-public-source research programme from Elite Careers Strategy on how highly selective employers and universities say they choose people. The current canonical corpus contains 1387 governed selection signals backed by 494 official source records. It is designed so a journalist, researcher or answer engine can move from a direct answer to the underlying evidence without guessing the denominator or scope.
Bank AI rules: 18 checked, stage-specific findings
In the 18-bank cut of WGI’s 67-employer AI Rules Register v2.1, observed 2026-10-23, official statements on AI in selection were located for 10 banks; none was located on the official pages checked for 8.
Of the 7 banks with coded rules on candidates’ own AI use, 5 allow help with written applications (with the limits recorded below), 5 prohibit AI in interviews and 0 explicitly allow it in interviews. These are stage-specific counts, not a blanket ban or permission across recruitment.
This is a purposive list, not a representative sample of banks. No statement located does not mean no policy exists. The research records published statements; it does not audit actual hiring practice, enforcement or AI detection. Market and programme scope differ by record. Snapshot observation date: 2026-10-23; media presentation revised 2026-10-06.
Current published bank tracker · Full 67-employer tracker. The original eight-institution AI comparison is a separate historical subset; its denominator must not be substituted for either of these.
Bank-by-bank evidence and official sources (10)
Open a bank for stage codes, official sources and scope. “Not stated” is a coding gap, not permission. Employer AI disclosures do not establish that candidates may use AI.
Barclays
WGI-AIR-011: full evidence record. Market: Global careers site. Programme: Early careers, apprentices, experienced hires. Observed: 2026-10-23.
- Written application: Not stated (no stage-specific statement coded)
- Online tests: Not stated (no stage-specific statement coded)
- Interviews: Prohibited · official source
- Employer’s own AI use: Not stated (no stage-specific statement coded)
BNP Paribas
WGI-AIR-017: full evidence record. Market: United Kingdom. Programme: UK Early Careers and Experienced Hire. Observed: 2026-10-23.
- Written application: Not stated (no stage-specific statement coded)
- Online tests: Not stated (no stage-specific statement coded)
- Interviews: Not stated (no stage-specific statement coded)
- Employer’s own AI use: States AI not used · official source
Official pages checked: source 1.
Citi
WGI-AIR-016: full evidence record. Market: Global. Programme: All applicants (careers privacy notice). Observed: 2026-10-23.
- Written application: Not stated (no stage-specific statement coded)
- Online tests: Not stated (no stage-specific statement coded)
- Interviews: Not stated (no stage-specific statement coded)
- Employer’s own AI use: Discloses own AI use · official source
Official pages checked: source 1.
Deutsche Bank
WGI-AIR-012: full evidence record. Market: Global. Programme: Students and graduates (video interview stage). Observed: 2026-10-23.
- Written application: Not stated (no stage-specific statement coded)
- Online tests: Not stated (no stage-specific statement coded)
- Interviews: Discouraged · official source
- Employer’s own AI use: Not stated (no stage-specific statement coded)
Official pages checked: source 1.
HSBC
WGI-AIR-006: full evidence record. Market: Global. Programme: Not stated (general recruitment). Observed: 2026-10-23.
- Written application: Allowed with limits · official source
- Online tests: General authenticity rule only · official source Listed as unacceptable use; no test-specific rule.
- Interviews: Prohibited · official source Listed as unacceptable use of GenAI.
- Employer’s own AI use: Discloses own AI use · official source
Macquarie Group
WGI-AIR-009: full evidence record. Market: Global template pages. Programme: Graduates and interns; all candidates. Observed: 2026-10-23.
- Written application: Allowed with limits · official source
- Online tests: Prohibited · official source
- Interviews: Prohibited · official source
- Employer’s own AI use: Not stated (no stage-specific statement coded)
Nomura
WGI-AIR-013: full evidence record. Market: Americas, EMEA, Asia ex-Japan, India early careers. Programme: Graduate and intern. Observed: 2026-10-23.
- Written application: Allowed with limits · official source
- Online tests: Prohibited unless authorised · official source
- Interviews: Prohibited · official source
- Employer’s own AI use: Not stated (no stage-specific statement coded)
Official pages checked: source 1.
Rothschild & Co
WGI-AIR-014: full evidence record. Market: Global. Programme: Students and graduates. Observed: 2026-10-23.
- Written application: Allowed with limits · official source
- Online tests: Restricted; exact rule not captured · official source
- Interviews: Restricted; exact rule not captured · official source
- Employer’s own AI use: Not stated (no stage-specific statement coded)
Official pages checked: source 1.
Standard Chartered
WGI-AIR-015: full evidence record. Market: Global. Programme: Early careers assessments. Observed: 2026-10-23.
- Written application: Not stated (no stage-specific statement coded)
- Online tests: Not stated (no stage-specific statement coded)
- Interviews: Not stated (no stage-specific statement coded)
- Employer’s own AI use: Discloses own AI use · official source Coding Simulation; the same FAQ says AI is not used to evaluate Workplace Scenario video responses or the Skills Insight Assessment.
Human involvement: the published score is compared with a threshold set in advance by human decision-makers. This does not establish that a human reads every application. Official source.
Official pages checked: source 1.
UBS
WGI-AIR-002: full evidence record. Market: Global. Programme: Not stated (all applicants). Observed: 2026-10-23.
- Written application: Allowed with limits · official source
- Online tests: Prohibited · official source
- Interviews: Prohibited · official source
- Employer’s own AI use: Not stated (no stage-specific statement coded)
Separate human-decision statement: observed 9 September 2026, not re-observed on 23 September. Official source. Do not treat it as freshly verified.
No statement located (8)
Goldman Sachs; JPMorganChase; Morgan Stanley; Bank of America; Jefferies; Lazard; Evercore; Centerview Partners. This finding is limited to the official pages checked at the snapshot date.
Download the bank-only evidence
18 banks, six stages each. Both formats retain scope, register dates, official quotations, source links and stage notes, including banks where no statement was located.
Bank-only evidence (CSV) · Bank-only evidence (JSON) · Reuse guide and limitations (TXT)
Citation and reusable data
Akram, H. (2026). What banks officially tell candidates about using AI. Who Gets In? AI Rules Tracker v2.1, bank cut (18 banks), observed 2026-10-23. https://whogetsin.com/trackers/ai-policy/banks. Credit: Who Gets In?, a research programme of Elite Careers Strategy. CC BY 4.0.
Full 67-employer register (JSON; not bank-only) · Full 67-employer register (CSV; not bank-only) · Methodology and coding rules · Corrections.
Employers allow AI help on the written application but draw the line at interviews
Short form. Of 31 selective employers that publish a rule on candidates' use of AI, 25 allow AI help with the written application and none forbid it; 18 forbid AI in interviews and none allow it.
Full form. Counts are of the 31 employers, out of 67 checked, whose official pages address candidates' own use of AI, observed 23 October 2026. "Allowed" includes allowed with limits. "Forbidden" includes forbidden unless the employer authorises it.
Universe. 67 selective employers in banking, asset management, private equity, law, consulting, the Big Four and prestigious UK graduate schemes, on a purposive WGI check list (not a random or complete sample). Observed: 23 October 2026.
What this does not show. That employers without a located statement allow AI, or have no policy. That any employer can detect AI use, or enforces its rule consistently. That the counts describe all selective employers (the list is purposive, not a sample). That a rule applies in every market or programme beyond the scope recorded for that employer.
Cite as. Akram, H. (2026). AI rules for candidates at selective employers: a source-verified register. Who Gets In? AI Rules Tracker v2.1, observed 23 October 2026. https://whogetsin.com/trackers/ai-policy
Open the tracker and employer lookup · JSON · CSV · Chart (SVG) · Citation objects for all 4 findings · Licence CC BY 4.0 · Report a correction
Everything needed to cite or challenge the research
Open the current journalist briefing
Current corpus, benchmark, flagship comparisons, limits and citation wording.
Download the complete source pack
Frozen + append-only canonical rulings, provenance, audit files and derived research objects. No private client evidence.
29 authoritative · 0 strong · 0 partial · 1 research gaps
The research is measured by hard questions, not page count.
Open citation.json
Preferred attribution, scope rule, machine routes and what must not be inferred.
80 words that describe the research accurately
Who Gets In? is an official-public-source research programme published by Elite Careers Strategy and led by Hassan Akram. It examines how highly selective employers and universities publish their selection processes across programmes and markets. Evidence is governed at institution, programme, geography, selection-feature and observation-date level, with denominators and limitations kept beside findings. The current corpus contains 1387 dated facts (signals) backed by 494 official source records. It is not a ranking, acceptance-rate database or representative prevalence study of all elite institutions.
Start from the object closest to the question
How do elite institutions actually select people?
The bounded category-level answer from the six-market foundation.
BlackRock across five priority hubs
Same-employer 2027 evidence across London, New York, Hong Kong, Singapore and GCC.
Morgan Stanley Singapore and Dubai
New official 2027 programme evidence resolved earlier public-source gaps while preserving the failed-search history.
Where can candidates use AI?
AI Rules Tracker v2.1: 67 employers checked, 37 official statements located, observed 23 October 2026. The original 8-institution subset remains valid for its own denominator.
Is there human review after digital selection?
Defined human-review subset N=7; scope preserved.
What has actually changed?
1 genuine comparable before/after pair currently meet the WGI standard. Each promoted change retains its earlier source, current source, comparable scope and reporting limit.
What Who Gets In? must not be reported as proving
- No market-wide or sector-wide prevalence unless a finding publishes an explicit denominator and inclusion rule.
- No ranking of institutions or markets, and no claim that one system is harder, fairer or more predictive.
- No hidden scoring rubric, university weighting, acceptance odds or universal conversion rate unless an institution itself publishes it and the claim is scoped to that evidence.
- “Not yet evidenced” and “research exhausted” describe the state of reviewed public evidence. They do not mean a real-world programme or practice is absent.
- Same-day research corrections, moved pages and newly discovered older material are not longitudinal employer change.
- Hassan Akram has been featured in the Financial Times. That is biographical credibility for Hassan/ECS, not FT validation, endorsement or publication of Who Gets In?.
- No private ECS client file is used as evidence in the public WGI research layer.
Preferred attribution
Suggested citation: Akram, H., Who Gets In? (Elite Careers Strategy), [finding, dossier or question title], [evidence/revision date stated on the object], [canonical URL].
Preserve the institution, programme, geography, denominator, observation date and limitation attached to the evidence. For machine use, see citation.json and benchmark-answers.json.
Hassan Akram
Hassan Akram founded Who Gets In?, a research project on how highly selective employers choose early-career talent. It is run by Elite Careers Strategy, the private career advisory firm he also founded. Across recruitment work and ECS advisory, he reports reviewing more than 10,000 applications and related candidate materials. He is accountable for WGI coding decisions, source governance, publication rulings and corrections.
He has delivered invited sessions for student and MBA communities connected with Yale School of Management, Harvard Business School and MIT Sloan. Those are community sessions, not institutional affiliations or endorsements. He has been featured in the Financial Times; the Financial Times did not publish or validate Who Gets In?.
Dated evidence on AI and early careers
Deloitte and PwC historical series distinguish hiring volumes, selection changes and employer statements about work. Population changes and causal limits remain explicit.
This page: revised . Corpus: 1387 dated facts (signals) from 494 official source records, last canonical update . Release notes
