Emmersion

Emmersion, Rosetta Stone's assessment division within IXL Learning, sells AI-scored adaptive TrueNorth speaking and writing tests and WebCAPE reading and grammar exams. They rate language proficiency on a 0-10 scale mapped to CEFR and ACTFL and are used for pre-employment screening, checking employees' language skills and placing students.
AssessmentScreening

HireAIScore rates Emmersion 44 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Emmersion ranks 49 of 82 vendors rated in Assessment AI, against a category average of 46. That score is drawn from 40 evidence items across 7 rubric criteria for Emmersion — 35 with a cited source, 5 recording that nothing was located.

Score breakdown

Emmersion, 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.

  1. Article 11 Technical Documentation

    Weight 20% · avg 50

    57

    +11.4

  2. Bias Audit Transparency

    Weight 18% · avg 41

    28

    +5.0

  3. FRIA Support

    Weight 15% · avg 31

    30

    +4.5

  4. Data Governance Disclosure

    Weight 15% · avg 53

    50

    +7.5

  5. Human Oversight Design

    Weight 12% · avg 55

    53

    +6.4

  6. Post-Market Monitoring

    Weight 12% · avg 38

    38

    +4.6

  7. Customer Documentation

    Weight 8% · avg 58

    60

    +4.8

Total weighted score

44.2(rounded to 44 for display)

Category avg is the mean raw score on that criterion across the 82 Assessment 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

Bias Audit Transparency

28raw / 100

Contributes 5.0 to the total. 28 against a 41 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 record

Evidence ledger

Items
40
Documentation
31
Public statement
4
Absence
5
With a source URL
35 of 40
Source hostnames
8
Fewest items
4
FRIA Support

These figures measure how thoroughly Emmersion was reviewed, not how Emmersion performed — an absence row, recording that nothing was located, is counted like any other item.

Article 11 Technical Documentation

5 items57
  • DocumentationCaptured Sep 25, 2026

    Emmersion's help centre publishes an Assessment Technical Reports folder: development and validation reports for its English (adaptive), Spanish, French, Italian, Japanese and Portuguese speaking tests, plus an elicited-imitation research summary, CEFR equivalency notes and proficiency-level descriptors.

    https://support.emmersion.ai/support/solutions/folders/60000475587

  • DocumentationCaptured Sep 25, 2026

    The English Form D report (2020 adaptive version) describes adaptive item selection, EAP ability estimation, scoring through a third-party speech-recognition API and an SDCA machine-learning model built from 260 examinees (Spearman .92 with the fixed form), but it has no limitations, intended-use or subgroup section.

    https://support.emmersion.ai/support/solutions/articles/60000694461-english-adaptive-truenorth-technical-report

  • DocumentationCaptured Sep 25, 2026

    The Spanish Form A report documents IRT calibration on 1,040 test records (729 kept after misfit screening) and validation against ACTFL OPIc ratings for 332 records (Spearman .854).

    https://support.emmersion.ai/support/solutions/articles/60000694456-spanish-truenorth-technical-report

  • DocumentationCaptured Sep 25, 2026

    US patent 11,488,489 (Adaptive language learning, assigned to Emmersion Learning and granted 1 November 2022) publicly describes the listen-and-repeat, automated scoring and adaptive item-selection method.

    https://patents.google.com/patent/US11488489B2/en

  • AbsenceCaptured Sep 25, 2026

    No model or system card, explainability statement, AI governance policy, ISO/IEC 42001 certificate or EU AI Act instructions for use was found on emmersion.ai, support.emmersion.ai, blog.emmersion.ai or Rosetta Stone's legal pages, and the writing test's promised technical paper is not in the technical-reports folder.

    Have a URL that disputes this?

Bias Audit Transparency

8 items28

FRIA Support

4 items30

Data Governance Disclosure

6 items50

Human Oversight Design

6 items53

Post-Market Monitoring

5 items38

Customer Documentation

6 items60

Editorial notes

Company overview

Emmersion was founded in 2015. It began as an online English-tutoring company in Rexburg, Idaho, later operated from Lehi, Utah, and was bought by IXL Learning in September 2022. Its website now calls it Rosetta Stone's expert assessment division and gives a San Mateo, California address, and its April 2026 security questionnaire names Rosetta Stone LLC as the vendor. The TrueNorth speaking test uses third-party speech recognition to score listen-and-repeat items and adjusts difficulty item by item. It also records open spoken answers that administrators can play back. The writing test pairs adaptive grammar and vocabulary questions with a timed written response. Results are given on a 0-10 TrueNorth scale mapped to CEFR and ACTFL, in nine speaking languages. Contact centres and multinationals use it to screen bilingual applicants. One published case study covers 1,700 candidates screened for Computacenter in Hungary, Poland, Spain, Mexico and South Africa.

Regulatory exposure

Emmersion's own help centre lists hiring and promotions or raises among the decisions customers make with its scores, and it recommends pass/fail cut-offs for high-volume screening. Employers using TrueNorth that way are likely running an automated employment decision tool under NYC Local Law 144 (enforced since 5 July 2023), which requires an annual independent bias audit and candidate notice. DCWP has promised tougher enforcement after a December 2025 State Comptroller audit, yet no audit of TrueNorth has been published. Hiring use in Hungary, Poland and Spain falls under EU AI Act Annex III point 4, and WebCAPE student placement touches point 3. The Digital Omnibus moved those high-risk obligations to 2 December 2027 without changing them. Because the tool scores accented speech, its sharpest exposure is national-origin discrimination. Illinois' amended Human Rights Act (from 1 January 2026) requires notice of AI use and bars AI use that has a discriminatory effect. California's FEHA automated-decision rules (since 1 October 2025) sit on top of existing rules that make accent- and English-proficiency-based decisions unlawful unless job-related, and they require four-year record retention. Colorado's SB 26-189 disclosure regime and California's CPPA ADMT rules both apply from 1 January 2027. Because the test is audio-based, Illinois' AI Video Interview Act probably does not apply.

Path to a higher score

The quickest gains would come from publishing evidence rather than claims. Emmersion should commission and publish an independent bias audit. It should give impact ratios by sex and race/ethnicity for LL 144 and, more importantly for a speech test, score gaps by first language, accent, age and recording device. It should either back up or withdraw the bias-free and 95%-accuracy marketing claims. It should replace the 2019-2021 technical reports with a current model card that names the speech-recognition engines, reports error rates across accents, and states limitations and intended use. Deployers need EU AI Act guidance, a FRIA template, NYC and Illinois candidate-notice text and advice on setting job-related cut-scores. On data, Emmersion should set and publish a fixed retention period for voice recordings. It should reconcile its conflicting privacy notices, controller and processor roles and the terms' claim to own recordings, and bring TrueNorth into Rosetta Stone's SOC 2 scope. It should also add admin audit logs that customers can see, a public status page, a security contact and a maintained model-version changelog.

Regulatory frame

What applies to assessment AI.

Assessment is the most-regulated surface in AI hiring. NYC LL 144, EEOC and OFCCP frameworks, Illinois HB 3773, and EU AI Act Annex III §4 all apply, with overlap on bias-audit obligations.

Keep reading

All assessment vendorsHow this score is builtAnnex IV section-by-section: what AI hiring vendors must documentWhat makes a defensible NYC LL 144 bias audit

Compare

Emmersion against its nearest-scoring peers.

The closest scores in Assessment AI, side by side.

Build any comparison

Ranking

Others rated in Assessment AI

Ranked 49 of 82 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.

  1. #46Clevry45F
  2. #47Cangrade45F
  3. #48Equalture44F
  4. #49EmmersionThis profile44F
  5. #50Prevue HR44F
  6. #51Retorio44F
All assessment vendors

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.