J
Jus Mundi
Jus Mundi is a research platform for international law and arbitration, built on a database of awards, orders, decisions, judgments, treaties, laws, case materials and institution profiles that the company states is updated daily and drawn from more than twenty institutional partnerships, with primary documents kept openly accessible through a free light version of its search engine. Its AI layer, Jus AI, launched in June 2023 and now described as agentic, answers research questions with citations to the database, assesses arguments against the facts and legal framework, compares cases across jurisdictions and institutions, drafts, translates and analyses, and shows each reasoning step so a user can check every citation and trace it to its source; a CiteMap feature maps citation links between documents, and the AI Enterprise plan connects Jus AI to a firm's own document database. The company also runs Jus Connect, a directory and conflict-check tool for arbitrators and counsel. Jus Mundi is a French simplified joint-stock company headquartered in Paris, founded in 2019, which closed a twenty million euro Series B in August 2024 and names Freshfields and White & Case among its customers. It publishes user terms, a privacy charter and per-user pricing, with individual academic access at twenty-nine euros a month, institutional academic access at two hundred and fifty euros, practitioners at one hundred and twenty-nine euros and firms on quotation, and states ISO 27001, ISO 42001 and SOC 2 certifications with its models served through Microsoft's Azure OpenAI Service and assistant interactions hosted on a separate locked database.
Capability grades
All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.
AI Centrality
How much of the product is actually AI. Whether the machine learning is the mechanism the buyer is paying for or a feature layered onto conventional software, and whether the vendor is specific about which is which.
The models are the engine of a core capability layered on a product that would still function without them. Jus AI answers, assesses, compares, drafts and translates over the database, and the vendor states it has trained models on its corpus since 2019; remove the models and the search engine, filters, interactive documents, CiteMap, Wiki notes and analytics remain, which is the research database the company sold before Jus AI launched in June 2023 and still sells on the Academic and Essentials plans without the assistant. Home page, pricing page and 2023 launch release read 6 September 2026.
Citation Accuracy and Hallucination Disclosure
Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.
Grounding is real and documented with linked primary sources, short of a testable accuracy figure. The home page states that Jus AI delivers fully cited answers from exclusive Jus Mundi data, that every reasoning step can be verified, each citation checked and each source traced to its origin, and the primary documents are the company's own database of awards, decisions, treaties and laws, kept openly accessible in a free light version so a reader can open what is cited. No accuracy figure, test set, benchmark or hallucination rate is published on the surfaces read, and the security page, which may carry more, could not be fetched. Home page and pricing page read 6 September 2026.
Autonomy and Oversight Model
What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.
A written commitment that the models work alongside a verifying lawyer with a real review surface, short of the full control structure. The vendor describes agentic legal reasoning that guides research with clarity and structure and exposes every reasoning step for verification, with citations traceable to source; the assistant answers and drafts for a practitioner to check. What is not published is a threshold at which any agentic step acts without review or a stated route back after a wrong answer beyond the user's verification. Home page and security-page excerpt read 6 September 2026.
Operational and Outcome Evidence
Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.
Testimonials and a client wall stand in for named deployments with figures on the surfaces read. The home page states that clients' experiences are the strongest signal and points to case stories about winning strategies and time saved, and a customer is quoted on the security page, unnamed in the excerpt, about confidence in uploading documents given ISO 27001 certification; the seed names Freshfields and White & Case as customers, which was not confirmed on a vendor surface read. The Meet Our Clients section and case stories were not opened and are the rebuttal route. Home page and pricing page read 6 September 2026.
Privilege and Confidentiality Posture
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
Confidentiality is addressed with two specific commitments and the rest of the picture could not be read. Segregation: the help centre states that assistant interactions are hosted on a separate and locked database, and the vendor states confidentiality is the foundation for sensitive casework. Third-party providers: the security page names Microsoft's Azure OpenAI Service and the enterprise plan connects Jus AI to a firm's own documents. Not established: any statement on training use, since the security page returned bot detection on fetch and the user terms were not read; any retention commitment; and any treatment of privilege or work product. The subscription agreement that governs firms is separate and not published. Home page, help centre excerpt and security-page excerpt read 6 September 2026.
UPL and Professional Responsibility Posture
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point.
No advice line or supervision statement was located on the surfaces read. The user terms describe the service as a research platform providing know-how and resources to international law and arbitration practitioners, and the product is sold to practitioners, firms, states and academics; whether the terms disclaim legal advice or address a practitioner's duties was not established because the terms were read only in excerpt. No jurisdiction limit beyond the subject-matter scope is named. User terms excerpt and home page read 6 September 2026; the full terms are the rebuttal route.
AI Governance and Bias Disclosure
Published governance over model behaviour: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
A published, independently audited governance framework, short of testing results or a named owner. The security page states that with ISO 42001 the company went further than security certification to show its AI is built responsibly and to the highest ethical standards, alongside ISO 27001 and SOC 2; the home page describes exposing every reasoning step as transparency. ISO 42001 is an audited AI management system standard and carries more weight than a self-published principles page; no test results, bias findings or accountable owner are published on the surfaces read, and the security page could not be fetched in full. Security-page excerpt and home page read 6 September 2026.
AI Safety and Data Stewardship
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
Some of the ground is covered and the surface that carries the rest could not be read. Stated: assistant interactions on a separate locked database, encryption, ISO 27001 and SOC 2 certification, Azure OpenAI as the model service. Not located on readable surfaces: a retention period, a deletion commitment, a sub-processor list or an incident-notification practice; the security page and trust centre returned bot detection on fetch and are the rebuttal route, together with the privacy charter, which was not opened. Home page, help centre excerpt and security-page excerpt read 6 September 2026.
AI Liability and Recourse
What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.
No liability position was located on the surfaces that could be read. The user terms, read in excerpt, define the services, materials and subscription structure without a located indemnity, cap, warranty or insurance position, and the subscription agreement that governs firms is stated in the Jus Connect terms to be a separate document, which is not published. This records what is locatable on the date and not a finding that no position exists; the full user terms and the subscription agreement are the rebuttal route. User terms excerpt and Jus Connect terms excerpt read 6 September 2026.
Practice Systems Integration Depth
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
Integrations are referred to without documentation an implementer could use on the surfaces read. The help centre headline describes an agent-to-agent integration bringing arbitration AI to where users already work, and the AI Enterprise plan connects Jus AI to a firm's own document database; nothing read names the systems, what syncs or in which direction, and no document management system is named. Help centre index and pricing page read 6 September 2026.
Deployment Model and Data Residency
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
Cloud delivery is stated and neither tenancy nor region is addressed on readable surfaces. The security page names Microsoft's Azure OpenAI Service as the model service and the help centre a separate locked database for assistant interactions; no region, residency option or tenancy model is stated in the material read, and the security page could not be fetched in full. Security-page excerpt and help centre excerpt read 6 September 2026.
Security Certifications and Trust Center
Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.
Certifications are real and stated on the vendor's own surface with a trust centre, short of a report reachable without asking. The security page states ISO 27001 and SOC 2 as proof of security and confidentiality by design and ISO 42001 for responsible AI, and points to a trust centre for the measures and certifications; a customer is quoted on the ISO 27001 certification. No certifying body, coverage period or report route is stated in the excerpt read, and the page and trust centre returned bot detection on fetch, so the access tier is not established and the lower tier is graded. Security-page excerpt read 6 September 2026.
Model Supply Chain Disclosure
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
The supply chain is partly disclosed. The security page states that Azure OpenAI Service enables Jus Mundi to offer its data security and that Microsoft makes AI available to the user community under its security and privacy controls, which names the provider and hosting arrangement; the 2023 launch release describes in-house machine-learning models trained on the company's database and a GPT-powered assistant. No specific model, inference region or change-notification commitment is stated on the surfaces read. Security-page excerpt and launch release read 6 September 2026.
Commercial Transparency
Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.
A figure, a unit and a term are published. The pricing page lists individual academic access at twenty-nine euros a month excluding tax, academic institutions at two hundred and fifty euros a month on a yearly subscription, and practitioners at one hundred and twenty-nine euros a month, each limited to one subscriber with monthly or yearly billing, with law firms, chambers and states on quotation, a seven-day free trial, and the features of each tier itemised; a newer plan structure of Academic, Essentials, AI Premium and AI Enterprise appears in search, and whether the AI plans carry a published figure was not established. Pricing page read 6 September 2026.
Firm and Practice Coverage
Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.
Segment and coverage are described with substance and the boundary is the product's stated scope. The buyer is international law and arbitration practitioners, law firms, chambers, states, public bodies and academics; the database is stated to cover all areas of international law and arbitration on any subscription, updated daily, with institutional partnerships and publisher content such as the ICC collection as add-ons; the vendor states the scope as international law and arbitration, which is the limit. No jurisdiction or matter type within that scope is named as unsupported. Pricing page and home page read 6 September 2026.
From €129 per monthEUR, as published, never converted
- A working lawyer pays 129 euros a month for one person, monthly or yearly.
- Students and academics pay 29 euros a month; a whole university pays 250 euros a month on a yearly deal.
- Law firms, chambers and governments have to ask for a quote.
- There is a free 7-day trial, and the original documents stay free to read in a light version.
- The AI assistant sits on higher plans whose price we could not confirm.
Published per-subscriber pricing in euros excluding tax on the pricing page: individual academic 29 per month, academic institution 250 per month on a yearly subscription, practitioner 129 per month, each limited to one subscriber, monthly or yearly billing, with law firms, chambers and states on quotation and a seven-day free trial; primary source documents remain free in a light search version. Search results show a newer tier structure of Academic, Essentials, AI Premium and AI Enterprise, and whether the AI plans carry a published figure was not established on 6 September 2026. No US dollar figure is published, so entryPriceUsd is left empty; the display carries the euro practitioner floor. Read 6 September 2026.
Legal Signals
What each signal meansA signal records what public sources say on the date shown. It is not a grade and it is not a recommendation. Where a signal reads Not addressed, it means the index did not locate the material in public sources on that date, which is a statement about disclosure rather than about the product.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
A published agreement or policy exists and none of it addresses the question either way, or the document that would answer it could not be read and the summary names the retrieval limit. The summary states which shape the silence takes: an improvement right granted that never names training, or no improvement right granted at all.
No located public material addresses whether customer content trains models. The help centre states that assistant interactions are hosted on a separate and locked database for privacy, which addresses storage rather than training; the security page, which may carry a training statement, returned bot detection on fetch, the user terms were read only in excerpt, and the subscription agreement governing firms is not published. Those three surfaces are the rebuttal route in either direction. Surfaces checked 6 September 2026.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
No located public material states how long prompts and outputs are retained.
No located public material addresses how long prompts to Jus AI or its outputs are retained. The help centre states interactions are held on a separate locked database without a period; the security page and privacy charter were not readable or not opened and are the rebuttal route. Surfaces checked 6 September 2026.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Segregation is asserted in public materials with no published detail on how it is enforced.
Segregation is claimed without documentation of a permission model. The help centre states that assistant interactions are hosted on a separate and locked database for additional privacy and security, which separates assistant data from the public research database rather than one customer's or matter's data from another's; the AI Enterprise plan connects a firm's own documents and nothing read describes how access within a firm is walled. Surfaces checked 6 September 2026.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
No located term or policy addresses third party requests for customer data.
No located public material addresses whether the customer is told when its data is demanded by a third party. The user terms were read only in excerpt, the subscription agreement is not published and the privacy charter was not opened; those are the rebuttal route. Surfaces checked 6 September 2026.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Sources are identified without stating the licence or rights basis.
The corpus is named and its update cadence stated, without the licence basis. The database is described as awards, orders, decisions, judgments, treaties, laws and case materials in international law and arbitration, updated daily, drawn from more than twenty institutional partnerships, with publisher content such as the ICC dispute resolution collection offered as add-ons and primary documents kept open in a free light version. The rights basis for the institutional and publisher content is not stated on the surfaces read. Pricing page, home page and 2023 launch release read 6 September 2026.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
The vendor computes and surfaces subsequent history itself, with the method described.
The vendor publishes its own citation signal. CiteMap, described in the user terms as part of the materials and in the launch release as a proprietary legal citation links creator, maps how awards and decisions cite one another across the database, and the home page states that Jus AI's citations can be checked and traced to source. Whether CiteMap flags subsequent negative treatment as a licensed citator would is not stated; it is a treatment signal of the vendor's own rather than a licensed one. Surfaces checked 6 September 2026.
Refusal and Uncertainty Behaviour
What does the product do when the answer is not in the corpus?
The vendor describes refusal or abstention behaviour in public materials.
The product documents how a user checks an answer rather than what the model does when it cannot answer. The home page states that every reasoning step is exposed and every citation traceable, which is a verification design; no abstention path or confidence signal for questions the database cannot support is described on the surfaces read. Recorded as documented on the strength of the exposed-reasoning design, with the abstention gap noted. Home page checked 6 September 2026.
Fabricated Citation Record
Does a public court record exist involving output from this product?
No court order, opinion or disciplinary record naming this product has been located as of the date shown. This is a statement about the public record, not a finding about the product.
No court order, opinion, arbitral decision or disciplinary record naming Jus Mundi or Jus AI was located as of 6 September 2026. The AI Hallucination Cases database maintained by Damien Charlotin was searched on both names together with a general search for court and tribunal findings; results returned sanctions involving general-purpose chatbots and the April 2026 Quebec Superior Court set-aside in ARIHQ c. Santé Québec, where an arbitrator's own award relied on fabricated authorities with no product named. This is a statement about the public record, not a finding about the product; a research tool that cites primary authority carries a real exposure on this signal, which is why the check was run on the tribunal record as well.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
No located public material engages with bar or ethics guidance.
No located public material names an ethics opinion, bar rule or professional responsibility framework. The pricing page offers expert advice and practical steps to guide teams through the AI shift, which is training rather than guidance alignment, and the user terms were read only in excerpt. Surfaces checked 6 September 2026.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials claim time savings without addressing billing or disclosure, and the product sits inside a fee relationship between a lawyer and a client where those savings would change the bill.
Law firms and chambers are named buyer segments and the published position on the bill is a savings claim: the home page frames client results as winning strategies and time saved, and research, drafting, translation and analysis completed in a fraction of the time. Nothing addresses how AI-assisted research is recorded or disclosed on a client's bill. Home page checked 6 September 2026.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A current subprocessor or model provider list is published.
The model provider is named on a public surface and a trust centre is stated to exist. The security page names Microsoft's Azure OpenAI Service as the service behind Jus AI and points to a trust centre for measures and certifications; whether the trust centre carries a full sub-processor list or a forwardable disclosure pack is not established because the page returned bot detection on fetch, which is why the row sits at this value rather than a higher one. Security-page excerpt checked 6 September 2026.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of the record are available, short of a document level export.
Some elements of a verification record are available and no export of a certification record is described. Every answer is cited to the database with reasoning steps exposed, so a practitioner can show the sources behind a research result; nothing states that a record of the model used, the sources retrieved and the human verification can be exported for a tribunal or court. Home page checked 6 September 2026.