Handshake
HireAIScore rates Handshake 50 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Handshake ranks 48 of 160 vendors rated in Screening AI, against a category average of 46. That score is drawn from 34 evidence items across 7 rubric criteria for Handshake — 27 with a cited source, 7 recording that nothing was located.
Score breakdown
Handshake, criterion by criterion
Each rubric criterion is scored out of 100 on the cited evidence, then weighted into the total.
Weighted contribution shown to the right of each bar.
Article 11 Technical Documentation
Weight 20% · category avg 49
48
+9.6
Bias Audit Transparency
Weight 18% · category avg 41
52
+9.4
FRIA Support
Weight 15% · category avg 31
30
+4.5
Data Governance Disclosure
Weight 15% · category avg 53
57
+8.5
Human Oversight Design
Weight 12% · category avg 56
58
+7.0
Post-Market Monitoring
Weight 12% · category avg 40
48
+5.8
Customer Documentation
Weight 8% · category avg 58
60
+4.8
Total weighted score
49.5(rounded to 50 for display)
Category avg is the mean raw score on that criterion across the 160 Screening AI vendors in scope of this rubric, this one included.
Strongest criterion
Customer Documentation
60raw / 100
Contributes 4.8 to the total. 60 against a 58 category average.
Weakest criterion
FRIA Support
30raw / 100
Contributes 4.5 to the total. 30 against a 31 category average.
Cited evidence
The evidence behind the score
Every score is backed by at least one cited piece of evidence. Open a criterion to see what was found, when, and where.
Download diligence recordEvidence ledger
- Items
- 34
- Documentation
- 23
- Audit report
- 3
- Public statement
- 1
- Absence
- 7
- With a source URL
- 27 of 34
- Source hostnames
- 6
- Fewest items
- 4
- FRIA Support
These figures measure how thoroughly Handshake was reviewed, not how Handshake performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
5 items48
- DocumentationCaptured Sep 25, 2026
The help-centre page lists each employer-side AI feature (job-form autofill, AI-drafted messages, audience selection, Job Matches and Talent List sorting, Pro applicant ranking) with its trigger, the employer's controls and the data it uses, plus governance principles such as keeping a human in the loop and internal model oversight.
- DocumentationCaptured Sep 25, 2026
This article explains how ranking works: criteria are generated from the job post, employers can use up to 10 of them (GPA, major, graduation year, location, AI-inferred skills or free text), starred criteria carry more weight, and applicants who meet many criteria go into a Recommended group.
- Audit reportCaptured Sep 25, 2026
Holistic AI's public LL 144 summary describes the AEDT but notes that the materials given to the auditor disclosed no model architecture, training data or performance metrics.
- DocumentationCaptured Sep 25, 2026
The Wolfia trust portal has an AI Features page that summarises the same features and links to the help article, but none of its roughly 40 documents or 10 badges is a model card, AI technical documentation or ISO 42001.
- AbsenceCaptured Sep 25, 2026
No model or system card, instructions for use, Annex IV-style technical pack, explainability statement, or reference to NIST AI RMF, ISO 42001 or the EU AI Act turned up in help-centre searches (AI, bias, fairness, EEOC, automated employment decision, AI Act), the trust portal, the joinhandshake.com legal pages or the sitemap.
Bias Audit Transparency
6 items52
- Audit reportCaptured Sep 25, 2026
The public Holistic AI NYC LL 144 summary for AI Applicant Management (distribution date 15 Sep 2025, audit expiration date 17 Sep 2026) shows standalone and intersectional impact ratios by sex and six race/ethnicity groups, with the lowest at 0.95 standalone and about 0.92 intersectional (Asian men), and offers no downloadable report.
- Audit reportCaptured Sep 25, 2026
The summary's data section says test data was used because no historical data with demographics was available: 3,600 synthetic CVs built from professionals' resumes in construction, finance and healthcare, each differing only by an injected sex and race label, were scored three times (10,800 scores), and a follow-up audit on real data is recommended.
- DocumentationCaptured Sep 25, 2026
Handshake says it runs independent AI bias audits where appropriate and links the Holistic AI page as its latest summary of results.
- DocumentationCaptured Sep 25, 2026
Applicant review was relaunched on Hiring Intelligence for all paid employers on 31 Jul 2026, after the audited version's distribution date, and no audit summary covering the relaunched system has been published.
https://updates.joinhandshake.com/announcements/handshake-employer-product-updates-august-28-2026
- DocumentationCaptured Sep 25, 2026
The trust portal lists NYC LL 144 as a public compliance document, but its link points to a legacy Holistic AI address that redirected to a login page when opened on 25 Sep 2026.
- AbsenceCaptured Sep 25, 2026
The help centre, trust portal, changelog and blog contain no fairness testing of employer-selectable criteria such as GPA, major or graduation year, and no audit of Hiring Intelligence sourcing, automated outreach or promotions targeting.
FRIA Support
4 items30
- AbsenceCaptured Sep 25, 2026
No EU AI Act, Article 26 deployer-duty or Article 27 FRIA material turned up in the help centre, the trust portal (AI Features page and document list), the legal pages (Access Terms, privacy policy, DPF notice) or the sitemap.
- DocumentationCaptured Sep 25, 2026
The privacy policy names Handshake Europe GmbH (Berlin) and Handshake International UK Ltd as the company's EU and UK representatives, which shows it processes data of EU and UK users.
- DocumentationCaptured Sep 25, 2026
The Handshake Access Terms (updated 20 Apr 2026) make the employer responsible for human oversight and for ensuring decisions made with the AI features comply with AI and data-protection law, which splits responsibility but gives the deployer no guidance.
- DocumentationCaptured Sep 25, 2026
The only AI compliance guidance for deployers is US-focused (writing EEOC-aligned criteria) and does not address fundamental-rights impacts.
Data Governance Disclosure
5 items57
- DocumentationCaptured Sep 25, 2026
The help article states that Job Matches, Talent List and applicant ranking use employer and job details plus candidates' interest and activity patterns, that protected categories such as race or gender are not used for relevance, that audience-selection data is pseudonymised, and that employer data is not shared with third parties for AI training.
- DocumentationCaptured Sep 25, 2026
The privacy policy (updated 3 Jul 2026) cites legitimate interests as the basis for AI-assisted features that use profile data, application materials and activity data, says resumes and screening answers are sent to third-party AI providers, and excludes private profiles from employer sourcing.
- DocumentationCaptured Sep 25, 2026
The trust portal lists 21 subprocessors with their purposes, including Anthropic, OpenAI and Google Gemini for AI workflows and Arize AI for model observability, along with SOC 2 Type II, CSA STAR Level 1 and Data Privacy Framework certifications; the SOC 2 report and data-retention policy are access-gated.
- DocumentationCaptured Sep 25, 2026
The Spring 2026 release says Hiring Intelligence matches candidates on skills, interests and engagement behaviour across hundreds of signals, learns from employers' thumbs-up/down feedback, and favours candidates likely to engage.
https://joinhandshake.com/blog/employers/product-release-employers-spring-2026/
- AbsenceCaptured Sep 25, 2026
No description of the ranking or matching models' training data, no analysis of proxy inputs such as GPA, school, graduation year or location, and no ISO 27001 or ISO 42001 listing was found.
Human Oversight Design
5 items58
- DocumentationCaptured Sep 25, 2026
Pro recruiters keep 3 to 10 criteria per job, star the ones that matter most and can edit them. Handshake rejects criteria it judges discriminatory, subjective or containing personal data, and recommended applicants get an AI summary and highlighted traits that explain the match.
- Public statementCaptured Sep 25, 2026
Handshake says it never makes hiring decisions for employers and keeps a human in the loop, and that employers can sort applicants by application date or name instead of the AI ranking.
- DocumentationCaptured Sep 25, 2026
Hiring Intelligence sourcing generates must-haves, nice-to-haves and criteria to avoid (used to screen candidates out) from the job post, and recruiters change them through free-text requests and refine matches by shortlisting or removing candidates.
https://support.joinhandshake.com/hc/en-us/articles/41268176962967-How-to-Source-Candidates
- DocumentationCaptured Sep 25, 2026
Once launched, automated outreach sends AI-personalised messages on the employer's behalf every weekday to candidates who match the AI-generated criteria, until the employer stops it or the run ends.
https://support.joinhandshake.com/hc/en-us/articles/15109901361815-Set-Up-Automated-Job-Outreach
- AbsenceCaptured Sep 25, 2026
No audit logs, per-jurisdiction settings (for example, for NYC), tooling for candidate AI notices or opt-outs, or route for candidates to contest a ranking is documented.
Post-Market Monitoring
4 items48
- DocumentationCaptured Sep 25, 2026
The Atlassian Statuspage has 11 components split between the US and UK/EU (students, EDU partners, employers, importer, reporting, plus the separate Handshake AI business) and a public incident history going back to 2021, but no postmortems and no component for ranking or matching.
- DocumentationCaptured Sep 25, 2026
A public LaunchNotes changelog, with dated employer updates back to 2023, records AI feature launches such as AI applicant management (8 Sep 2025) and Hiring Intelligence applicant review (31 Jul 2026), but only at the product level.
- DocumentationCaptured Sep 25, 2026
The trust portal names Arize AI as a subprocessor for model observability on AI recommendations and offers a security@ route for vulnerability reports, while its incident-response, logging and monitoring, and responsible-disclosure policies are gated.
- AbsenceCaptured Sep 25, 2026
No public dashboard for AI performance or bias, no model-version notes and no channel for reporting AI harms or contesting outcomes was found on the help centre, trust portal, status page or changelog.
Customer Documentation
5 items60
- DocumentationCaptured Sep 25, 2026
This guidance on writing EEOC-aligned AI matching criteria gives employers examples of what to use and avoid (no age, birthplace or culture-fit criteria) and reminds them that anti-discrimination law applies, but it never mentions GPA.
- DocumentationCaptured Sep 25, 2026
The Colorado Job Application Fairness Act resource tells employers to avoid school-year or graduation-date qualifications, but the AI-criteria guidance does not carry this over and offers a graduation-date cutoff as a compliant example.
- DocumentationCaptured Sep 25, 2026
The August 2026 release notes explain Hiring Intelligence, automated outreach and applicant management, list which plans get each, and state that employers make every hiring call.
https://joinhandshake.com/blog/employers/product-release-employers-bts-2026/
- DocumentationCaptured Sep 25, 2026
The student-facing AI explainer covers job recommendations and resume tools but does not tell job seekers that employers may rank or source them with AI.
- AbsenceCaptured Sep 25, 2026
No public DPA and no NYC LL 144, Illinois, California or EU AI Act guidance or candidate-notice template for employers was found, and the legal hub lists only the terms of service, Access Terms, privacy, cookie and DPF notices.
Editorial notes
Company overview
Handshake (legal entity Stryder Corp., San Francisco, founded 2014) runs an early-career recruiting network that connects students and recent graduates with employers. On the employer side, AI ranks applicants against criteria the employer sets. This launched as AI applicant management for Plus and Pro customers in September 2025 and was relaunched as Hiring Intelligence applicant review for all paid employers on 31 July 2026. The AI also finds and messages matching candidates and targets job promotions. Pro costs $450 a month and Enterprise pricing is custom. Handshake's separate AI data-labeling business is outside this assessment.
Regulatory exposure
Handshake has had its applicant ranking audited under NYC Local Law 144, so it treats the feature as an AEDT. The published Holistic AI summary, however, used synthetic CVs, lists an audit expiration date of 17 September 2026 that has now passed, and predates the July 2026 relaunch. DCWP has promised tougher enforcement after a critical December 2025 State Comptroller audit. Handshake's EU and UK affiliates act as its GDPR representatives. If its ranking, sourcing or targeted promotions are offered to EU employers, they fall under the EU AI Act's Annex III high-risk category, with obligations deferred to 2 December 2027. Employers using it in Illinois have owed AI-use notices and non-discrimination duties since 1 January 2026. Criteria such as GPA cutoffs or graduation-year windows can produce adverse impact by race or age. Colorado's SB 26-189 disclosure regime starts on 1 January 2027. California's FEHA automated-decision rules, including four-year record retention, have applied since 1 October 2025, and the CPPA's ADMT rules follow on 1 January 2027.
Path to a higher score
Publish a renewed LL 144 audit that covers the current Hiring Intelligence applicant review, using real applicant data where possible, and fix the trust portal's broken audit link. Test the criteria employers can choose (GPA, major, graduation year, location) for adverse impact and tell employers how to use them safely. Publish a model or system card for ranking and matching that covers training data, performance and limits. Give employers public NYC, Illinois, Colorado, California and EU AI Act guidance with notice templates, plus a public DPA. Add audit logs and tools for candidate notice and opt-out. Update the student AI explainer so job seekers know employers may rank and source them with AI.
Regulatory frame
What applies to screening AI.
Treated as high-risk under Annex III §4 when the screening output is decisive. EU AI Act Article 13 transparency obligations apply; deployers must give candidates a way to contest.
Compare
Handshake against its nearest-scoring peers.
The closest scores in Screening AI, side by side.
- Side-by-sideHandshake vs Indeed50F
- Side-by-sideHandshake vs Xobin50F
- Side-by-sideHandshake vs Paycom50F
- Side-by-sideHandshake vs Test Partnership50F
Ranking
Others rated in Screening AI
Ranked 48 of 160 by weighted total under rubric v1.0. The ordering is arithmetic on the rubric and carries no view on which tool suits a given hiring process.
- #45Xobin50F
- #46Paycom50F
- #47Test Partnership50F
- #48HandshakeThis profile50F
- #49Versant by Pearson49F
- #50Jobma49F
Conflicts of interest
No vendor pays for placement, scoring, or removal. Casework - the consulting firm that operates this directory - provides paid services to some vendors. Any active or recent (within 24 months) commercial relationship is disclosed on the affected vendor profile and the review is reassigned to an independent reviewer. See the full policy on About.
Casework has no commercial relationship with this vendor.