Genie AI vs Paxton AI: how they compare in 2026
Genie AI and Paxton AI both sell lower cost AI to small firms, solo practice and lean in house teams, and both publish a price. The difference between them is paperwork. Genie names its certificate to its number, issuing body and audit dates. It publishes a register of 35 controls and a subprocessor list of fifteen third parties with processing locations. It gives the customer an intellectual property indemnity with a stated remedy ladder, and it sets out its own regulatory position in terms, including that it is not regulated for legal services and that no legal professional privilege attaches. Paxton AI answers with the one thing Genie does not have: measured accuracy on public academic datasets. It reports 93.82 percent average on the Stanford Legal Hallucination Benchmark and 94 percent on CaseHold for its own citator, published with precision, recall and F1 by case type.
At a glance
All 15 axes, side by side
The same grid applied to every vendor in the index, graded from public sources. 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 drive the core capability, but they sit on a document product that would function without them. Drafting, playbook review, redlining, comparison across documents and tabular extraction are all generative, and the Eidetic Intelligence architecture is what the vendor sells against general purpose LLMs. Underneath sits a conventional platform: a docx editor, PDF import, document sharing with privacy controls, department spaces and a library of more than 500 templates. All of these appear on the Free plan, and Genie shipped the templates as an open source library in December 2018, before the assistant existed. Remove the models and a template repository with an editor remains.
The models are the product. Research, drafting, document analysis, medical chronologies and the citator are all generative or machine learning capabilities, and there is no underlying practice system that would function without them.
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.
Accuracy is measured and published, but as comparative scores rather than citation accuracy. The benchmarking program carries two dated head to head reports: 18 February 2026 against Claude Sonnet on a regulatory scenario across several jurisdictions, and 22 April 2026 against Claude CoWork on a commercial supply agreement. They score 15 and 10 dimensions respectively, with rationale per metric that the vendor says a reader can reproduce or critique. Those are comparative scores on Genie's own rubric, scored by Genie, rather than a measured citation accuracy figure or a hallucination rate with the test set described. Three headline marketing numbers match neither report and are not published on the benchmarks index. They are 92 percent accuracy rated by lawyers, a 140 percent advantage over GPT and Claude, and 90 percent on a 65 document study against 79.3 percent for Claude CoWork and 37.3 percent for ChatGPT. Genie's home page states that the architecture prevents hallucinations before they affect the final result, an unmeasured claim. The benchmarks page also carries the meta description 'Independent head-to-head benchmarks' while its body says Genie's own engineering team publishes them.
The disclosure is substantive, with a real gap. The vendor built its own citator and publishes precision, recall and F1 tables for it. It publishes measured accuracy of 93.82 percent on average on the Stanford Legal Hallucination Benchmark, broken out by task including case existence, citation retrieval and court identification. It also publishes 94 percent on the Stanford Casehold dataset of 2,400 examples for the citator. Both are public academic datasets, so an outsider can in principle run them again. Two gaps remain. The published results date from 2024, and no republication against the current product was found. The AI Confidence Indicator is a confidence signal rather than an explicit path for declining to answer.
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.
Autonomy is claimed prominently and the oversight mechanism is not described. Marketing sells agents that draft, review and negotiate contracts autonomously, invites the buyer to control when they need to and delegate when they do not, and describes agents handling complex deals across several documents and borders. What is published about the review point sits in the agreement rather than the product: clause 3.4 disclaims any guarantee that the software comprehends what is put into it, clause 5.9 states that recommendations, suggestions or actions taken by the software are not legally binding, clause 6.10 prohibits use for automated decision making with legal or similarly significant effects, and clause 21.7 instructs the user to have a human review each output as appropriate. Those are allocations of responsibility, not a described control structure, and no threshold, review surface or escalation path is published on the home page, the use case pages or the model training page. The direction of the gap is worth noting: the marketing sells delegation while the agreement prohibits the automated decision making that delegation implies.
The product is user directed rather than agentic, and the vendor states plainly across the site that it provides self help services at the user's specific direction and cannot give advice, which is an oversight posture of a kind. The AI Confidence Indicator is a real review aid. What is not published is a described control structure: what the system does on its own, what a lawyer must approve, and what happens after an output is wrong.
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.
Named customers with named individuals and at least one figure are published, without dates or method. The customer stories index carries attributed video case studies. Jude Legg, Compliance Lead at Firefish, speaks to cutting contract review time by more than 50 percent; Jafeth Bulsink, Head of Sales at Bright Renewables, to 25 salespeople negotiating contracts with no legal department; Neil Rowe, Head of Commercial at Cambridge United, to player signings; and Nazmus Tareque of AnthroTek to closing a seed round without a lawyer. A logo strip of 13 named organizations, including hyperexponential, Blockworks and the Jamaica Stock Exchange, sits alongside them, and the model training page states that Clifford Chance, Pinsent Masons and Withers ran pilots from 2018. No deployment date, scope or measurement method is published.
An unlabeled customer logo strip and three attributed testimonials from named individuals at named organizations stand in for deployment evidence. No figures, no dates, no described method, and no case study were found. A 2025 Legalweek Leaders in Tech Law award and a 22 million dollar Series A are recognition and funding rather than production evidence. Individual logos in the strip were not corroborated.
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.
The published commitment and the published agreement point in opposite directions on training, and the agreement is what a customer is held to. The security center states that customer documents, prompts and confidential data are never used to train shared or internal AI models, and the model training page is headlined on never training on your contracts. Clause 7.3 of the product terms grants Genie a nonexclusive, irrevocable, worldwide, perpetual license to use, copy, display, distribute and reproduce a customer's Documents to improve the software and, in terms, its algorithms and models. Clause 6.8 lets Genie use Your Data and Documents to make Improvements, and clause 7.5 vests ownership of those Improvements in Genie. The model training page's FAQ narrows its promise to contracts never training models that other customers see. On the other side, OpenAI and Anthropic are named and operate under zero data retention agreements, a DPA and a subprocessor list are published, and clause 20.7 says communications are not covered by legal professional privilege. No tenant or matter level segregation model and no retention period is published.
Confidentiality is asserted in general terms, and the documents behind it point the other way. The home page describes a secure closed model prioritizing the privacy of legal inquiries. The sitewide footer states that communications with the vendor are not protected by attorney client privilege or as work product. The terms of service require a user to warrant that submissions do not constitute confidential information, in a section scoped to questions, comments, feedback and other information about the Services rather than to uploaded documents. The privacy policy is scoped to website and account information, lists no document or prompt data, and states that the vendor does not process sensitive information, on a product that takes in medical records. No training prohibition, retention period for uploaded documents or segregation by matter was found in any of them.
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. Where the advice line is not the duty a product raises, the axis is read through the nearest professional duty it does raise: judicial conduct rules and the reviewing duty for products sold only to courts, and the duty to bill for time actually spent for products that draft time entries.
A specific published position on the advice line exists, without supervision duties or other jurisdictions. Clause 20 of the product terms states that Genie AI Limited is not a law firm and is not authorized or regulated by any body regulating legal services. It cannot provide reserved legal services, including litigation, probate, notarization and administering oaths, and operates only in general commercial matters. It then sets out the consequences for the buyer: no requirement to hold professional indemnity insurance, no right to complain to a legal ombudsman, no access to a compensation fund such as the SRA's, and no legal professional privilege over what the user receives. Clause 6.7 excludes domestic, personal and noncommercial use, so there is no consumer product requiring disclosure. Nothing published addresses a lawyer's own competence and supervision duties. The whole treatment is drawn from the England and Wales regulatory perimeter, on a product sold across more than 150 jurisdictions including the United States. No equivalent statement for any of them is published.
Paxton publishes a real position on the advice line in a footer on every page. It states that Paxton is not a law firm and not a substitute for an attorney, that communications are not protected by attorney client privilege or as work product, and that it gives no advice, explanation, opinion or recommendation about legal rights, remedies, defenses, options, selection of forms or strategies. Jurisdiction is addressed in part: the terms state the service is not intended for use where that would be contrary to local law. Nothing published addresses a lawyer's competence and supervision duties. The self help framing sits oddly against marketing aimed at attorneys, so it is not clear who the disclaimer is written for.
AI Governance and Bias Disclosure
Published governance over model behavior: 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 described quality mechanism exists where a governance position should be. The model training page sets out a five layer account in which qualified lawyers contribute training data, set prompt engineering standards, supervise retrieval logic and rate every release against a panel benchmark before it ships. The security center publishes a specific EU AI Act position: Genie is not a high risk system under the Annex III categories, the obligations applying to it came into force on 2 August 2026, and it maintains AI literacy measures for staff operating the system. No individual or role inside the vendor is named as accountable for model behavior, no prerelease testing regime is published beyond accuracy benchmarking, and nothing is published on uneven output across matter types, counterparties or populations. The ISO 27001 certification covers information security, a different subject.
No governance position for model behavior was found: no named internal owner, no testing regime before release, no responsible AI statement, and nothing on uneven output across matter types, parties or populations. The trust center lists security control families, which is a different subject. The white papers index was not reviewed item by item and may say more.
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.
Substantive published policy covers almost all of this ground, missing a retention period. The security center publishes a register of 35 controls grouped as an auditor would walk them. They include AES-256 at rest and TLS 1.2 or higher in transit, quarterly access reviews recorded in an Access Rights Review Log, a joiner, mover and leaver process, environment segregation, peer reviewed code with deployment separated from development, and tested encrypted backups. A business continuity plan carries a recovery time objective of one hour for critical services, and incident response is documented, with incidents tracked through an Incident and Corrective Action Log. Breach practice is stated: notification to affected customers without undue delay and, where Genie is controller, a report to the UK Information Commissioner's Office within 72 hours. A current subprocessor list names 15 third parties with the purpose, processing location and security measures for each. The register states only that formal procedures govern the retention and secure disposal of customer data, and no retention period for documents or prompts is published.
A generic privacy policy covers the website without addressing what happens to documents and prompts after processing. The security page does publish real controls: encryption in transit and at rest, quarterly access reviews on a least access principle, vendor risk management and a security development lifecycle, and the open trust center lists control families. But the privacy policy is scoped to account and marketing data, its retention section speaks only to the life of an account, and the terms say only that certain transmitted data is maintained to manage performance of the service. No retention period for uploaded documents, no subprocessor list, and no incident or breach notification practice is published.
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.
A full allocation of loss is published, with nothing covering the output itself. Clause 18.2 caps aggregate liability at the fees paid or payable in the 12 months before the first event giving rise to the claim, and clause 2.4 sets a separate cap of one thousand pounds for the Free Package. Clause 9 gives the customer a defense and indemnity against third party claims that use of the software infringes intellectual property rights. Conditions sit in 9.2, a remedy sequence in 9.3 of procuring a license, modifying the software or terminating, and six named exceptions in 9.4, including Your Data and any use under a Free Package. There is no recourse when the output is wrong. Clause 5.8 excludes liability for reliance on the software's outputs, recommendations or actions, clause 21.3 disclaims any warranty that content is accurate or complete, and clause 21.4 excludes all liability for reliance. Clause 20.4 states that Genie is not required to hold professional indemnity insurance, and it makes no claim to hold any.
Liability is addressed only through a standard limitation clause that disclaims the exposure the product creates. The published terms disclaim all warranties, including any representation about the accuracy or completeness of content, and cap total liability at the amount paid in the six months before the claim. They require the user to indemnify the vendor and state that the vendor has no liability for loss or corruption of transmitted data. No indemnity running to the customer, warranty on output or insurance position was found. The allocation of loss is published, and a buyer can read it before signing.
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.
Integration categories are named, and no documentation an implementer could use is published. Genie's home page lists MCP connectors, an API, and native integrations with email, CRM, cloud storage and electronic signature, and the Enterprise plan adds SSO alongside API and MCP integration. A Word add in has its own page. Every 'Learn more' link on the home page integration card resolves to an anchor rather than a destination, and no developer index, integration directory or help center documentation is published from the site navigation or the footer. No document management integration such as iManage or NetDocuments, matter management, electronic billing or court filing connection is named. That fits a product sold to commercial teams rather than firms, but leaves the depth unverifiable.
No integration into document management, matter management, contract lifecycle or filing systems was found in vendor material, and the platform navigation lists none. A third party review reports a Microsoft Word add in; that is not vendor material. The product is positioned as a browser based tool for solo and small firm users.
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.
Residency is published in detail, and the tenancy model is not stated. The security center states that the platform runs on Amazon Web Services with customer documents stored in UK and EU data centers. The subprocessor table gives a processing location for each of the 15 third parties, which separates where data sits from where it is processed: AWS and MongoDB in the UK and EU, CloudAMQP in the EU, and OpenAI and Anthropic in US and EU data centers. Nothing published states whether the platform is multi tenant or single tenant, and no private or on premises option and no region choice for the customer is published. The same subprocessor table lists Google Cloud and Stripe at global data centers, which is broader than the UK and EU statement on the security page.
The service is hosted, and the tenancy model is not stated. The terms of service state that the service is hosted in the United States and that a user accessing it from elsewhere consents to transfer and processing there, which is a real residency statement. Not found: any tenancy model, regional option, single tenant or private deployment, or statement of where processing happens as distinct from where data is stored.
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.
Certification is real, current and documented in detail, with the report itself held back. The security center names the standard as ISO/IEC 27001:2022, gives certificate number 310012019 and names the issuing body as Citation ISO Certification. It publishes the original approval date of 7 February 2019, a current certificate dated 7 February 2026, expiry on 6 February 2029, a last external audit on 29 April 2026 and a last management review in June 2026, with the page marked last reviewed 30 July 2026. The vendor states plainly that it does not hold a SOC 2 report. The certificate, the ISMS scope and the Statement of Applicability are available on request under NDA through a named security mailbox with a stated response within two working days. They are not offered as a download, so the ISMS scope is withheld. The control register refers to independent penetration testing after significant changes but names no testing partner and publishes no summary.
Badges appear with no scope, no date and no report available. There is an open trust center that anyone can reach. It lists SOC 2, ISO 27001 and HIPAA as compliant and GDPR as in progress, and names security control families. It carries no attestation report, coverage period, audit scope or auditor name. The security page says the vendor rigorously adheres to these standards, which is adherence rather than attestation. The pricing page describes the platform as HIPAA certified, though HIPAA has no certification scheme. Its annual plan card cites ISO 20701, which is not an existing standard, while the monthly card on the same page cites ISO 27001. Separately, section 1 of the terms of service states that the services are not tailored to comply with industry specific regulations including HIPAA, and that a user whose interactions are subject to such laws may not use them. The product sells medical chronology and medical billing features alongside its marketing claim of HIPAA compliance.
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 providers are named, and the change commitment runs the wrong way. OpenAI and Anthropic are both listed on the published subprocessor table with their purpose, processing locations in US and EU data centers and stated security measures. The security center adds that both operate under Zero Data Retention agreements, so inputs and outputs are not retained by the provider after processing. The model training page states that Genie passes legal context at inference time instead of fine tuning, specifically so newer foundation models can be swapped in as they land. No specific model or version is identified for either provider. On notification, the subprocessor page says changes will be posted there and asks readers to check back regularly, which leaves noticing a supply chain change to the customer.
No model provider is named, no architecture is described, no subprocessor list exists, and there is no commitment to notify customers when the supply chain changes. The vendor refers to operating within a secure closed model without identifying what sits underneath.
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.
Real pricing is published across the whole range, including a starting price for the enterprise plan. The pricing block on the home page carries a figure, a unit and a term for all three plans. Free is $0 for one user with limited tokens and no document exports, Pro is from $75 per month for one user with enough usage for roughly five documents, and Enterprise is from $600 per month for unlimited users and unlimited tokens. Each plan's contents are itemized, and signup needs no sales call and no credit card. Every figure carries a 'From', so each is a starting price rather than a rate. Nothing published states what onboarding, the dedicated customer success manager or the bespoke AI training program bundled into Enterprise add. Prices display in US dollars while clause 13.3 of the product terms states that payment will be made in pounds sterling unless stated otherwise.
Real pricing is published for part of the range, and Enterprise pricing is withheld. The pricing page publishes a figure, a unit and a term: $499 per user per month, or $2,999 per user per year billed annually, with the plan contents itemized. Signup is self serve with a 7 day free trial, and the terms of service disclose the exact card authorization hold of $499. The Enterprise plan is custom and based on volume, with no figure, floor or seat minimum published. What implementation adds is not stated for either plan.
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.
Coverage is documented precisely on the dimensions the vendor sells, and the limits are stated in the agreement. Dedicated pages exist for six user groups including in house lawyers, procurement, sales, finance and founders, for startup, mid market and enterprise companies, and for six industries: energy, construction, technology, real estate, mining and sport. The Enterprise plan states coverage of more than 150 jurisdictions and ten languages. Clause 20.2 of the product terms names what Genie cannot do, reserved legal services including litigation, probate, notarization and administering oaths, and confines the product to general commercial matters. Clause 6.7 excludes domestic, personal and noncommercial use. No law firm segmentation is published, nothing addresses government or court use, and the practice dimension is expressed as industry verticals rather than areas of law, so the sectors served are known but not which law is covered.
Segment and practice coverage is described with substance. Named segments are solo attorneys, small and midsize firms and in house teams, with dedicated pages for personal injury, family law, employment law, criminal law and corporate law. The research coverage page names every federal and state database included by jurisdiction and court, which tells a buyer where research coverage stops. Not found: any treatment of government or court use, and any statement of which firm sizes or practice areas the product is not built for.
The 12 legal signals, side by side
Recorded rather than graded. These are the questions a practitioner has to answer before a tool touches a client matter, and the answers are taken from public material only.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
The published agreement permits what the marketing says never happens, with no requirement to deidentify. Clause 7.3.2 of the product terms grants Genie an irrevocable perpetual license to use a customer's Documents to develop and improve its algorithms and models, and clause 6.8 adds that Genie may use Your Data and Documents to make Improvements. The security center and a dedicated model training policy both state that customer documents, prompts and confidential data are never used to train shared or internal AI models.
Where the marketing and the agreement disagree, the agreement governs. The model training page narrows the promise the same way, limiting it to contracts never training models that other customers see, and says the 200,000 businesses using Genie contribute new legal context through normal use. No opt out from the clause 7.3.2 license is offered.
Read the terms of service dated 27 Feb 2026, the privacy policy dated 20 Feb 2026 and the security page on 29 Aug 2026. No located term or policy addresses training on customer content either way. The contribution license clause grants the vendor rights to access, store, process and use information the user provides, following the privacy policy, and the privacy policy is scoped to website and account information and does not mention uploaded documents, prompts or model training. Home page language about a secure closed model is not a training commitment.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Retention is acknowledged in the security center control register without a stated period, and the same register states that on contract termination customer data can be returned or securely deleted in line with the customer agreement. Neither the security center nor the product terms publishes a retention window for documents, prompts or outputs, no customer configuration of a window is described, and no zero retention option is offered at the Genie layer. Zero data retention is stated for the third party model providers rather than for Genie's own storage.
The terms of service acknowledge that data transmitted to the service is maintained for the purpose of managing performance, with no period stated and no customer control described. The privacy policy retention section speaks only to personal information for the life of an account and does not address documents or prompts. No zero retention option was located as of 29 Aug 2026.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Separation is asserted at the organization level in the home page security section. It is supported by features described as document sharing with privacy controls, business department spaces on the Pro plan, and a role model of Workspace Owner, Administrator and Authorized User in which the Workspace Owner sets access rights. How that separation is enforced is not published. The home page, the security center, the subprocessor page and the product terms show no description of tenancy separation, no access control model for retrieval, and no statement of whether one department space can be reached from another.
The buyer here is an in house or commercial team rather than a firm, so the question is separation between customers rather than walls between matters.
Searched the home page, platform pages, security page, the open trust center, the terms of service and the help center index on 29 Aug 2026. No vendor material addresses ethical walls or matter level segregation. No document management system integration was located either, so there is no source system permission model for retrieval to inherit. The security page documents least access controls for vendor personnel, which is a different question.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Clause 22.6 of the product terms permits disclosure of Confidential Information where required by law, a governmental or regulatory authority or a court, on condition that the disclosing party, to the extent legally permitted, gives the other party as much notice as possible and takes account of its reasonable requests about the content. The commitment sits in the mutual confidentiality clause rather than in a dedicated government or law enforcement request clause, so it reaches customer material through the definition of Confidential Information rather than by naming customer data directly. No transparency report is published.
Read the terms of service and the privacy policy in full on 29 Aug 2026. Neither contains a government or law enforcement request clause. The privacy policy section on when and with whom information is shared names only business transfers such as a merger or acquisition. No commitment to notify a customer of a third party request was located, and no transparency report exists.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
The model training page identifies the source material by composition and origin: more than 300,000 contracts and 10 million clause revisions built since 2017, plus open source templates, legislation and case law spanning the US and UK legal systems and described elsewhere as 150 jurisdictions. It names pilots with Clifford Chance, Pinsent Masons and Withers from 2018 and continuing contributions from the 200,000 businesses now using the product.
No individual database or source is named for the legislation and case law, no license or public domain basis is stated for any component, and no update cadence or lag is published.
The vendor publishes a research coverage page that names its source databases individually rather than by jurisdiction alone. It lists the United States Code, the Constitution, the Code of Federal Regulations, every federal circuit and the Supreme Court, and six named administrative tribunals. For each state and territory it lists the statutes, administrative code, court rules and named courts. No license or public domain basis is stated for any of it, and no update cadence or lag is published.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
Nothing on the home page, the use case pages, the model training page, the benchmarks index or in the product terms addresses whether authority is checked for subsequent history, and no citator, treatment signal or good law indicator is described. The product drafts and reviews contracts rather than retrieving primary law for research, so a citator is not part of what it sells, although its corpus is stated to include legislation and case law and one published benchmark scores a legal authority citations dimension.
The vendor built and surfaces its own treatment signal, the Paxton AI Citator, described as patent pending, which reports whether a case has been overruled, affirmed, questioned or cited by later cases and returns explanations in Bluebook format. The method is described and quantified: evaluated against the Stanford Casehold dataset of 2,400 examples with published precision, recall and F1 figures by case type. It is the vendor's own signal rather than a licensed commercial citator such as Shepard's or KeyCite. The published evaluation dates from 2024 and no republication against the current product was located.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
No explicit path for declining to answer or abstaining is documented, and no confidence or grounding score is described. The nearest published material points elsewhere: the home page describes a system gated on quality that validates outputs at every stage and prevents hallucinations before they impact the final result, which is a claim about prevention rather than abstention, and the model training page says qualified lawyers review where the model should defer, which describes a design review rather than published product behavior. Clause 3.4 of the product terms disclaims any guarantee that the software comprehends what is put into it.
The product exposes an AI Confidence Indicator, announced alongside the vendor's benchmark results and described as helping a lawyer judge the reliability of a response. Searched the site, platform pages, benchmark posts and help center index on 29 Aug 2026 and located no explicit no answer or abstention path, and no published statement of what the product does when it cannot ground an answer.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
The AI Hallucination Cases database maintained by Damien Charlotin, which tracks decisions worldwide where a court addressed hallucinated AI content and records the tool implicated where known, records no court order, opinion or disciplinary record naming Genie AI or the company Genie AI Limited, and 2026 sanctions summaries in the trade press name neither. The product drafts and reviews contracts rather than producing court filings, so its output does not commonly take the form of citations in a brief.
One court record mentions this product, and the court's findings address the lawyer's failure to verify rather than the product's output. In Flycatcher Corp. Ltd. v. Affable Avenue LLC (S.D.N.Y., No. 24 Civ. 9429, opinion and order of 5 February 2026), counsel told the court that after running arguments through vLex he put his motion brief through Paxton AI as a cite checker and then through a further AI check. He said the cite checker introduced citation errors that he took for corrections; the court found that explanation not credible, found that he never cite checked the brief himself, struck the filings and entered default judgment against his client.
The court made no finding about Paxton AI's own output. The AI Hallucination Cases database maintained by Damien Charlotin was searched on 2 October 2026 and returns this matter as the only record naming the product.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Clause 20 of the product terms engages with professional responsibility in specific terms but does not name ethics guidance. It states that Genie is not a law firm and not regulated for legal services, identifies reserved legal activities it cannot perform, and names the consequences for the buyer, including no legal ombudsman route and no access to a compensation fund such as the SRA's. The home page, the model training page, the security center, the benchmarks index and the legal pages show no engagement with any named ethics opinion.
That includes ABA Formal Opinion 512, US state bar guidance, and Law Society or SRA guidance on the use of AI. The material describes Genie's own regulatory status rather than the obligations its buyers are bound by.
Searched the home page, platform and practice area pages, the news and insights index, the white papers index and the webinars page on 29 Aug 2026. No engagement with any named ethics opinion was located, including ABA Formal Opinion 512 and state bar guidance. The sitewide footer carries a specific unauthorized practice disclaimer, which is a statement about the vendor's own role rather than engagement with the guidance its buyers are bound by.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials are framed around time and cost saved. They include a customer quote on cutting contract review time by more than 50 percent, another on closing a funding round without a lawyer, and marketing copy on intelligence greater than a law firm at a fraction of the price. The home page, the pricing block, the use case pages and the legal pages show no record of AI assisted work per matter for fee purposes and no guidance on billing, fee or client disclosure treatment.
The product is sold to business teams rather than to firms billing a client, so the fee question arises differently, but nothing addresses it either way.
Vendor material is framed around time saved, describing streamlined document analysis to save time and accelerate decisions, with testimonials quoting time savings. Searched the site, platform pages, help center index and legal pages on 29 Aug 2026 and located no per matter record of AI assisted work intended for fee purposes, and no published guidance on billing, fee or client disclosure treatment.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A current subprocessor list is published without a gate, naming 15 third parties with the purpose of processing, the location and the security measures for each, and identifying OpenAI and Anthropic as the model providers that see content. A standard data processing agreement including the EU Standard Contractual Clauses and the UK International Data Transfer Addendum is published alongside it, as is a model training policy.
What is not published is material for clients that a firm or department could forward to satisfy an AI clause: there is no consent or notification pack, and the ISO 27001 certificate, ISMS scope and Statement of Applicability are available only on request under NDA.
The open trust center publishes compliance badges and security control families with no gate, which is a real access route. What a firm would need to satisfy a client AI clause is absent: searched the trust center, the security page, the terms of service and the privacy policy on 29 Aug 2026 and located no subprocessor list, no statement of which model providers see customer content, and no client facing consent or notification material. There is also no data processing agreement published.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of a record exist as a byproduct of how the product works. Genie's home page states that it edits, redlines and manages negotiations directly in tracked changes and that the result exports to Word, so what the AI changed in a document is attributable and portable, and a version comparison feature shows what changed between documents. Two elements are missing: the model used is not disclosed anywhere on the property, and no export designed to cover model, sources retrieved and human verification together is described. The product generates contracts rather than court filings, so a judicial standing order is not its usual context.
Some elements of a disclosure record are available. The product returns citations with the relevant sources highlighted, and the citator produces a treatment result with a written explanation, so sources retrieved and their status can be shown per answer. Two elements are missing: the model used is not disclosed anywhere on the property, and no per document export covering model, sources and human verification together was located as of 29 Aug 2026.
Which one fits
Choose Genie AI if
- Your reviewer wants the certificate, not the badge. Genie AI names the standard as ISO/IEC 27001:2022, gives certificate number 310012019 and the issuing body as Citation ISO Certification, publishes the original approval date, the current certificate date, the expiry, the last external audit on 29 April 2026 and the last management review, and states plainly that it does not hold a SOC 2 report.
- A client asks which third parties touch its contracts. Genie AI publishes a subprocessor list naming fifteen third parties with the purpose, processing location and security measures for each, identifies OpenAI and Anthropic as the model providers operating under zero data retention agreements, and states that customer documents are stored in United Kingdom and European Union data centers.
- You need to know what you give up by buying from an unregulated provider. Clause 20 of Genie AI's product terms states that it is not a law firm and not regulated for legal services, cannot perform reserved legal activities including litigation and probate, and then works through the consequences: no professional indemnity requirement, no legal ombudsman, no compensation fund and no legal professional privilege.
Choose Paxton AI if
- You want accuracy measured on something an outsider could re run. Paxton AI publishes 93.82 percent average on the Stanford Legal Hallucination Benchmark broken out by task, including case existence, citation retrieval and court identification, and 94 percent on the Stanford CaseHold dataset of 2,400 examples for its citator, with precision, recall and F1 figures published by case type.
- You need to know whether a case still stands. Paxton AI built its own citator, reporting whether a case has been overruled, affirmed, questioned or cited by later cases with explanations in Bluebook format, and publishes a research coverage page naming its sources database by database, including the United States Code, the Code of Federal Regulations, every federal circuit and, for each state, the statutes, administrative code, court rules and named courts.
- You want to buy a seat without a sales process. Paxton AI publishes 499 dollars per user per month or 2,999 dollars per user per year billed annually, itemizes what the plan includes, sells through self signup with a seven day free trial, and discloses the exact card authorization hold in its terms.
In summary
Genie AI
Genie AI is a contract drafting, review and negotiation platform sold to business teams rather than law firms. In house legal sets the playbooks that commercial, procurement and sales colleagues work within as they draft from templates, review third party paper, redline in tracked changes and run multi document workflows. According to the AI Legal Index, its strongest published material is documentary. It holds ISO/IEC 27001:2022 and names the certificate number, issuing body and audit dates. It publishes a register of 35 controls and a subprocessor list naming fifteen third parties with processing locations, and it prices all three tiers. It publishes no retention period, no AI governance material and no accuracy figure for generated output.
Paxton AI
Paxton AI is a self serve legal AI assistant aimed at solo attorneys, small and midsize firms and in house teams. It covers research, drafting, document analysis, medical chronologies and medical billing summaries, and it sells through self signup rather than a sales process. According to the AI Legal Index, Paxton publishes measured accuracy on public academic datasets: 93.82 percent average on the Stanford Legal Hallucination Benchmark and 94 percent on CaseHold for its own citator. It publishes a price of 499 dollars per user per month. No model provider, training position or AI governance material is published.
Questions buyers ask
Genie AI vs Paxton AI: which is better for a small firm?
Genie AI publishes far more paperwork. It names its certificate to its number and dates, lists fifteen subprocessors and gives an indemnity with a remedy ladder. It also gives a full account of its regulatory position. Paxton AI publishes the one thing Genie does not, which is measured accuracy on public academic datasets. Both publish a price. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
How much do Genie AI and Paxton AI cost?
Both publish figures. Genie AI prices three tiers on its home page: Free at zero for one user, Pro from 75 dollars per month, and Enterprise from 600 dollars per month for unlimited users, each marked from, so each is a floor rather than a rate. Paxton AI publishes 499 dollars per user per month or 2,999 dollars per user per year with a seven day free trial and self signup, with only its enterprise tier withheld. Neither states what implementation adds. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
Does Genie AI train its AI on your contracts?
Two documents answer differently, and the AI Legal Index records the agreement. Genie AI's security center and its model training policy state that customer documents, prompts and confidential data are never used to train shared or internal AI models. Clause 7.3.2 of the product terms grants Genie an irrevocable, perpetual, worldwide license to use a customer's documents to develop and improve its algorithms and models. Clause 6.8 adds that Genie may use customer data and documents to make improvements. No opt out is published.
What does Paxton AI publish about security certifications?
An open trust center lists SOC 2, ISO 27001 and HIPAA as compliant and GDPR as in progress, and names security control families, with no attestation report, coverage period, audit scope or auditor published. Two documents describe HIPAA differently: the pricing page calls the platform HIPAA certified, while section 1 of the terms of service states that the services are not tailored to comply with industry specific regulations including HIPAA and that users whose interactions are subject to such laws may not use them. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
What do Genie AI and Paxton AI both leave unpublished?
Neither publishes an AI governance position: no accountable owner, no pre release testing regime and nothing on uneven output across matter types or populations. Neither states a retention period for documents, prompts or outputs. Neither documents what the product does when it cannot ground an answer, although Paxton exposes a confidence indicator. And neither names an ethics opinion, including ABA Formal Opinion 512. From the AI Legal Index, based on each vendor's own published materials as of September 3, 2026. No vendor pays for placement.
Neither vendor paid for inclusion, placement or a grade, and neither reviewed this page before it published. Everything on it comes from public material on the dates shown. How the index grades.
Two things need both documents named. On Genie AI, the security center and a dedicated model training policy state that customer documents, prompts and confidential data are never used to train shared or internal AI models. Clause 7.3.2 of the product terms grants Genie an irrevocable perpetual license to use a customer's documents to develop and improve its algorithms and models, with no de identification qualifier and no opt out. The agreement is the document that governs, so the AI Legal Index records it. On Paxton AI, the published benchmark results date from 2024, and no republication against the current product is published. A figure from two model generations ago is weaker evidence than the same figure published this year. Neither vendor reviewed this page.