Product and intended use
Define the configuration, people, information and tasks being assessed.
Product, documents and plan covered
Public claim only
The legal hub published at 27 September 2026 sets out the document set: a Master Services Agreement (Source) and a separate Master Services Agreement for Managed Service (Outsource), Terms of Use, Referral Terms, a Privacy Statement, a Data Processing Addendum, a sub-processor list, a GDPR page, a Schedule 5 Artificial Intelligence Addendum and a Security Policy, plus Australia-specific Compliance Programme Addendum and Standard Operating Procedure PDFs. This profile covers the Source (self-serve platform) MSA and the Schedule 5 AI Addendum. The Managed Service (Outsource) arrangement is a separate agreement and is not covered.
Ask: Will our Order be on the Source MSA or the Managed Service (Outsource) MSA, and will it expressly reference the Schedule 5 AI Addendum?
Deployment model and regional instances
Public claim only
MSA clause 21 defines Service as First AML's proprietary cloud service as identified in the relevant Order. No on-premises or self-hosted option appears in the MSA or on any page read. Two separate application instances are published in the site navigation: an ANZ login at app.firstaml.com and a UK/EU login at app.eu.firstaml.com. The documents read do not say where either instance is hosted.
Ask: Which application instance would an Australian practice use, and in which country and cloud region does that instance store case data and run processing?
AI features in scope and Embedded AI
Public claim only
Four AI features are named and separately described on the AI feature page checked 27 September 2026: Entity Insights, which searches public records, registries and credit data to produce a structured open-source report on an entity; an AI document reader that reads trust deeds, constitutions and complex documents, extracts beneficial owners and certification details and builds the ownership structure inside the case; AI-supported individual verification, which captures a live video alongside the ID document and uses facial biometrics to confirm the person submitting matches the document; and advanced AI screening, where an AI screening agent reviews and labels each result as a false positive, false match or true match for the user to accept or override. The Schedule 5 AI Addendum adds a second, distinct category: Embedded AI, defined as AI used to structure, normalise, enrich, extract or otherwise prepare data in the course of providing the Services, which does not itself determine, recommend or propose a decision and is not subject to opt-out.
Ask: List every AI Feature and every Embedded AI use that would be active on our tenant, and state for each whether it is on by default.
Jurisdictions and regulatory scope
Public claim only
Separate Compliance Programme Addendum and Standard Operating Procedure documents are published for New Zealand, Australia and the United Kingdom on the legal hub. The Schedule 5 AI Addendum clause 3.5 states that Output does not constitute legal, regulatory or compliance advice and that using AI Features does not discharge any obligation that applies to the customer, including under applicable anti-money laundering and counter-terrorism financing laws. MSA clause 9.3 states First AML does not warrant that use of the Service allows the customer to meet any legal obligations it may have regarding the retention or storage location of verification, accounting, taxation or transaction records.
Ask: Which Australian AML/CTF rules and which state conveyancing requirements does the Australian Compliance Programme Addendum map to, and on what cycle is it updated?
Vendor content directing how AI describes the product
Not confirmed
Checked 27 September 2026 for a page or block telling AI models, assistants or crawlers how to describe the product. No such content was found on the AI feature page, the Trust Deed Reader page, the legal hub, the Master Services Agreement (Source), the Schedule 5 AI Addendum or the sub-processor page. The site navigation and footer on those pages list no /ai-info, /for-ai or llms.txt path. The check does not cover the whole firstaml.com domain, which has regional variants for Australia, New Zealand, the United Kingdom and a global default and was not mapped in full.
Ask: Do you publish any page, block or file that sets out how AI assistants, crawlers or large language models should describe First AML, and if so where is it?
Data, controls and requirements
Compare the actual configuration with firm, client and jurisdiction requirements, supported by evidence.
Client authority, notices and consents
Public claim only
MSA clause 4.1 makes the customer responsible for its Customer Data, including content and accuracy, and has the customer represent and warrant that it has made all disclosures and received all consents needed for the rights granted in clause 3.1, without infringing any law, third-party right or the privacy policies that apply to that data. Schedule 5 AI Addendum clause 3.3.2 adds that if the customer uses AI Features to process personal information it must comply with all applicable data privacy laws, including providing privacy notices and obtaining consent where required. Clause 3.3.1 makes the customer solely responsible for its Inputs. No model notice or consent wording is supplied by First AML in the documents read.
Ask: What client notice and consent wording do you expect an Australian practice to have in place before a trust deed, a constitution or an identity document is put through the AI document reader or AI-supported individual verification?
Foundation-model providers
Public claim only
Schedule 5 AI Addendum clause 1 defines AI Providers as the third-party model providers used in the AI Features and Embedded AI, as described in the sub-processor list. The sub-processor list checked 27 September 2026 names one generative-AI entry: OpenAI LLC, purpose 'Generative AI Solutions', country United States. The same list names Amazon Web Services, Inc (Cloud services, United States) and Google LLC (Cloud services provider, United States). The list states a country for each entity but does not say which feature uses which provider, or in which region a provider processes data.
Ask: Which model does each of Entity Insights, the AI document reader, AI-supported individual verification, advanced AI screening and Embedded AI call, which legal entity provides it, and in which region is the endpoint?
Sub-processor register
Public claim only
A sub-processor page is published and was read on 27 September 2026. It lists 37 third parties, each with a purpose and a country, across cloud services, data analysis, identity verification, communication, customer support and sales and marketing, plus three First AML group entities that may act as sub-processors to provide support and similar services: First AML Limited (New Zealand), First AML Pty Limited (Australia) and First AML UK Limited (United Kingdom). The page is not dated and sets no notice period: it says only that sub-processors may change over time and encourages readers to check back regularly, with a privacy@firstaml.com contact for questions. Neither the MSA nor the AI Addendum sets a notice period or an objection right for new sub-processors.
Ask: What advance notice will we get before a sub-processor is added or changed, and do we have a right to object, given the published list sets no notice period and the MSA contains no sub-processor change clause?
Training on customer content
Public claim only
Schedule 5 AI Addendum clause 3.1 states the AI Providers are prohibited from using Inputs or Outputs to train or improve their models for their own purposes and process Inputs and Outputs on a zero-retention basis, and that First AML will not train or fine-tune any underlying third-party large language model used by the AI Features using Inputs or Outputs. Clause 3.6 then permits First AML to use Data, Inputs and Outputs for Evaluation and Validation, defined as testing AI Features to measure accuracy, reliability, safety and performance and expressly stated not to constitute training. Clause 3.7 permits First AML to derive and use Aggregated Insights from Data, Inputs or Outputs for operation, improvement, security, fraud-prevention and analytics purposes. MSA clause 12 separately allows First AML to collect logs and utilisation statistics and to aggregate that data with Customer Data in non-personally identifiable form for analytics, improvements and benchmarking.
Ask: Under clause 3.6, who inside First AML can read our case documents, Inputs and Outputs for Evaluation and Validation, and can we opt out of clause 3.6 and clause 3.7 for our tenant?
Feedback, telemetry and aggregated use
Public claim only
MSA clause 11 states that if the customer provides feedback or suggestions regarding the Service, First AML may use them without restriction or obligation, and that modifications and improvements to the Service belong to First AML. MSA clause 12 allows collection of logs, utilisation statistics and other technical data about the customer's use, and aggregation of that data and Customer Data in non-personally identifiable form. Schedule 5 AI Addendum clause 3.7 permits Aggregated Insights, stated not to identify any specific customer and not to be treated as Customer Content. No in-product rating control for AI Features is described in the documents read.
Ask: If a user rates an AI output in the product, where does that rating go, and can the analytics and Aggregated Insights permissions in MSA clause 12 and AI Addendum clause 3.7 be switched off for us?
Retention of case data, prompts and outputs
Public claim only
Schedule 5 AI Addendum clause 3.1 states the AI Providers process Inputs and Outputs on a zero-retention basis. MSA clause 3.4 states that after the Subscription Term and the Off-boarding Period in clause 10.3, First AML may delete Customer Data in accordance with its standard schedule and procedures; that schedule is not set out in the MSA and no number of days appears. MSA clause 9.3 states First AML does not warrant it will maintain Customer Data without loss and is not obliged to review it for accuracy. Clause 10.3 permits Customer Data to remain in standard backups after deletion. The Security Policy, which MSA clause 3.2 incorporates by reference, was not opened within the credit budget.
Ask: State the standard schedule referred to in MSA clause 3.4 as a number of days, separately for case files, uploaded identity documents, uploaded trust deeds, AI Inputs and Outputs, screening results and audit logs.
Deletion and written confirmation
Public claim only
MSA clause 10.3 gives a 30-day Off-boarding Period after expiry or termination during which, on written request, First AML will make Customer Data available for export or provide a copy in a commonly used machine-readable format and give reasonable transition assistance. After that period First AML will delete Customer Data in accordance with clause 3.4 and will, on written request, confirm deletion in writing, except where law requires continued retention. The same clause allows Customer Data and Confidential Information to be retained in standard backups after deletion, subject to the confidentiality restrictions. No period is stated for the written confirmation or for backup expiry.
Ask: Within how many days of our written request will you confirm deletion in writing, and how long do the standard backups referred to in clause 10.3 retain our case data and identity documents after deletion?
Vendor and group access to content
Public claim only
The published sub-processor page states that First AML Limited (New Zealand), First AML Pty Limited (Australia) and First AML UK Limited (United Kingdom) are group entities that may act as sub-processors to provide support and similar services to the customer. Schedule 5 AI Addendum clause 3.6 permits First AML to use Data, Inputs and Outputs for Evaluation and Validation, which implies staff access to that material but does not describe who may do it or under what approval. MSA clause 3.2 says First AML uses reasonable technical and organisational measures as described in the Security Policy; that Security Policy was not opened within the credit budget, so no staff-access, approval or logging control has been verified.
Ask: Which First AML staff and which of the three group entities can open an Australian practice's case files, identity documents and AI Outputs, under what ticketed approval, and can we obtain those access logs?
Access control, credentials and AI settings
Public claim only
MSA clause 2.2 restricts access to Users, requires each User to keep login credentials confidential and not share them, makes the customer responsible for each User's compliance and for actions taken through their accounts, and requires the customer to notify First AML promptly of any credential compromise. Schedule 5 AI Addendum clause 2 states that some AI Features may be enabled by default and that the customer may disable AI Features at any time through the AI settings in the platform, while Embedded AI is not subject to opt-out. Neither document read mentions single sign-on, SCIM provisioning, enforced multi-factor authentication or role-based permissions.
Ask: Is single sign-on with enforced multi-factor authentication available on our plan, and which administrator role controls the AI settings that enable or disable AI Features for other users?
Encryption in transit and at rest
Not confirmed
MSA clause 3.2 states First AML uses reasonable technical and organisational measures designed to protect the Service and Customer Data as described in the Security Policy, and clause 21 defines Security Policy as the current version published by First AML. The Security Policy is published at https://www.firstaml.com/about/help/security/ and was not opened, because the four-operation credit budget was spent on the legal hub, the Master Services Agreement, the Schedule 5 AI Addendum and the sub-processor list. No algorithm, protocol version or key-management statement was found in the documents read.
Ask: State the encryption algorithms and protocol versions for case data at rest, identity documents and biometric material, data in transit, and the path to OpenAI LLC, and say whether customer-managed keys are available.
Independent security assurance
Not confirmed
A Trust Centre is published at https://trust.firstaml.com/ and is linked from the footer of every First AML page read, and a Security Policy page is published on the legal hub. Neither was opened within the four-operation credit budget. No certification, audit report or attestation is named in the Master Services Agreement, the Schedule 5 AI Addendum or the sub-processor list, and the MSA contains no clause requiring First AML to hold or maintain any named security standard.
Ask: Provide the current SOC 2 Type II report or ISO/IEC 27001 certificate with its scope and date, and state whether the four AI Features and Embedded AI are inside that scope.
Contracting entity, data-protection terms and liability
Public claim only
The Master Services Agreement (Source) read on 27 September 2026 is expressed to be between 'First AML and the customer identified in the Order' and names no First AML legal entity. Clause 20.2 makes the governing law, jurisdiction and venue whatever the Order sets out. Clause 20.5 states the agreement comprises the MSA, Orders, SOWs, the Policies (defined in clause 21 as the Security Policy and the Privacy Policy), any AI Addendum referenced in an Order, and the DPA if applicable under clause 3.3. Clause 3.3 engages the DPA only where processing falls within the scope of the European General Data Protection Regulation or the UK Data Protection Act 2018, so on the terms read the DPA is not engaged by Australian Privacy Act processing alone. Clause 13.2 caps each party's liability at amounts paid or payable in the prior 12 months, with a Super Cap of three times that amount for Special Claims, defined in clause 21 as unauthorised disclosure of Customer Data caused by a First AML breach.
Ask: Which First AML entity will sign our Order, which law and courts will it name, and since clause 3.3 engages the DPA only for GDPR or UK Data Protection Act processing, what contractual data-protection terms will apply to our Australian Privacy Act personal information?
Cross-border transfers and Australian residency
Public claim only
The published sub-processor list gives a country for each sub-processor: the United States for Amazon Web Services, Inc, Google LLC, OpenAI LLC, Datadog, Snowflake, Fivetran and others; the United Kingdom for ComplyAdvantage, Equifax, Creditsafe, Armalytix and Companies House; Ireland for Kyckr and Know Your Customer; Austria for Kompany; Singapore for Onfido PTE Ltd; Switzerland for Tresorit AG; New Zealand for Centrix; and Australia for Frankie Financial Pty Ltd, SFDC Australia Pty Ltd and Sisense Australia Pty Ltd. The list states the country of each entity and does not say where any of them stores or processes data. No document read states a hosting region for the ANZ instance, and no Australian data-residency commitment and no statement about where inference for the AI Features runs was found. Australian residency is therefore recorded as unknown rather than inferred from the separate app.firstaml.com instance.
Ask: For an Australian customer, in which country is case data stored and backed up, and in which country does inference for each AI Feature and for Embedded AI run? Is Australian-only storage and Australian-only inference available, and on what terms?
Incident notification
Not confirmed
No obligation on First AML to notify the customer of a security incident or eligible data breach was found in the Master Services Agreement (Source) or the Schedule 5 AI Addendum read on 27 September 2026. The only notification obligation found runs the other way: clause 2.2 requires the customer to notify First AML promptly of any compromise of a User's login credentials. Clause 3.3 engages the Data Processing Addendum only where GDPR or the UK Data Protection Act 2018 applies, so for Australian Privacy Act processing alone the DPA notification terms are not engaged on the terms read. A status page is published at status.firstaml.com; it was not opened. The Data Processing Addendum PDF and the Privacy Statement were not opened within the credit budget.
Ask: What contractual notification period from discovery applies to an Australian customer for a security incident or eligible data breach, including one at a sub-processor such as OpenAI LLC, and which signed document carries it?
Change notice for features, terms and models
Not confirmed
MSA clause 19 says First AML will give reasonable prior notice, by email or through the Service, of any new feature that materially changes the customer's use and of any optional Paid Feature and its fees, that changes under clause 19 will not materially decrease overall functionality during a Subscription Term, and that any other amendment or modification to the agreement must be agreed in accordance with clause 20.6. Schedule 5 AI Addendum clause 2 says something different for the AI Terms: First AML may change, suspend or discontinue any AI Features or Embedded AI at any time, may impose limits on or restrict access to all or part of the AI Features without liability, and the customer's continued use of AI Features following the posting of any changes to the AI Terms constitutes acceptance of those changes. The two mechanisms differ: the MSA requires agreement to amendments, and the AI Addendum treats continued use after posting as acceptance. The sub-processor page separately sets no notice period and asks readers to check back regularly.
Ask: Which mechanism governs changes to the AI Terms in our agreement, clause 20.6 of the MSA or acceptance by continued use under clause 2 of the AI Addendum, and what written notice will we get before a model provider or a sub-processor changes?
Export and exit
Public claim only
MSA clause 3.4 lets the customer export Customer Data during the Subscription Term using the export features described in the Documentation. Clause 10.3 gives a 30-day Off-boarding Period after expiry or termination during which, on written request, First AML will make Customer Data available for export or provide a copy in a commonly used machine-readable format and provide reasonable transition assistance. Clause 10.2 allows either party to terminate for an uncured material breach after 30 days written notice, insolvency or cessation of operations. The clauses do not name the export formats, do not say whether AI Outputs, screening results, audit logs or extracted ownership structures are included, and do not allow the Off-boarding Period to be extended.
Ask: What export formats are produced, do they include AI Outputs, screening labels, extracted ownership structures and audit logs, and can the 30-day Off-boarding Period in clause 10.3 be extended in our Order?
Third-Party Platforms and connectors
Public claim only
MSA clause 6 states that use of Third-Party Platforms is subject to the customer's own agreement with the relevant provider and not the MSA, that First AML does not control and has no liability for them including their security, functionality, availability or how they use Customer Data, and that if the customer enables a Third-Party Platform First AML may access and exchange Customer Data with it on the customer's behalf. Clause 5 allows First AML to disable integrations with Third-Party Platforms if the customer breaches clause 2.3 or clause 4. Clause 21 defines Third-Party Platform as any platform, add-on, service or product not provided by First AML that the customer elects to use. No named connector or integration appears in the documents read, and no administrator approval control for enabling one is described.
Ask: Which Third-Party Platforms would be enabled on our tenant, what Customer Data does each receive, and can enabling a new one be restricted to a named administrator with our written approval?
Retrieval from sources outside the tenant
Not confirmed
Two of the four AI features are described as reaching outside the tenant. The AI feature page states Entity Insights 'searches public records, registries and credit data to produce a structured open-source report on any entity, with follow-up questions to go deeper', and that advanced AI screening 'runs alongside standard adverse media screening to surface court proceedings, criminal filings and content behind paywalls'. The sub-processor list names registry, credit and media suppliers consistent with that. Neither the Master Services Agreement nor the Schedule 5 AI Addendum states whether general open-web retrieval is used in addition to those named suppliers, and no setting to limit retrieval to named sources is described. This is a design a firm must accept or configure.
Ask: Do Entity Insights and advanced AI screening perform general open-web retrieval, or only queries against the named registry, credit and media sub-processors, and can retrieval be restricted to a named source list for our tenant?
What the AI can do without a user action
Public claim only
Schedule 5 AI Addendum clause 2 states that some AI Features may be enabled by default, that the customer may disable AI Features at any time through the AI settings in the platform, and that Embedded AI forms part of the Services and is not subject to opt-out. Clause 1 defines AI Features as producing an Output that determines, recommends or proposes a decision, classification, assessment or course of action for an authorised user to review, and defines Embedded AI as processing that does not itself determine, recommend or propose a decision and does not carry out profiling or automated decision-making producing legal or similarly significant effects on an individual. Clause 3.2 states the customer is responsible for determining where Outputs require human review before being accepted or acted upon. The AI feature page describes the screening agent labelling each result for the user 'to accept or override', and a marketing heading on that page states cases 'progress themselves' against compliance rules and review steps set by the customer's team.
Ask: Which AI Features are on by default on a new Australian tenant, and can any of them close a case, clear a screening hit, accept an extracted ownership structure or pass a verification without a user action?
Human review before an output is used
Public claim only
Schedule 5 AI Addendum clause 3.2 states AI Features are designed to support, not replace, human judgement, that the customer is responsible for configuring how AI Features are used in line with its own risk assessment including determining where Outputs require human review before being accepted or acted upon, and that the customer and its authorised users retain control over and responsibility for any resulting decision or action. Clause 3.4.1 makes the customer solely responsible for the use of Outputs and for ensuring they are true and accurate. Clause 3.5 states Output does not constitute legal, regulatory or compliance advice and that use of AI Features does not discharge the customer's AML/CTF obligations. The Addendum places the review obligation on the customer and does not describe any product control that enforces it.
Ask: What product controls can force a named reviewer to sign off an extracted ownership structure, a screening label or a biometric verification result before the case can progress, and can those controls be mandated firm-wide?
