A
AccuFind Redact
AccuFind Redact reads every document in a freedom of information or public records request and flags the pages and people that may be exempt, each under a named provision, so officers can decide what to release. It comes from AccuFind Pty Ltd of Adelaide, the maker of AccuFind Virtual Assistant, and is sold to government FOI, records and release teams.
An officer starts a case by describing the request, the applicant, key people and any pending proceedings, and Redact uses those answers on each page. It takes in PDFs, Office files, emails, mailboxes, archives and scanned pages, sets aside duplicates and blank pages with a reason, and matches each person across the case. Smart Flags each ask one question about one provision, such as confidential sources, legal professional privilege or personal privacy, and mark the result Strong, Possible or Weak.
Officers apply, reject or dismiss every flag, and a suggestion never redacts by itself. Sealing confirms every page has been reviewed and every redaction has a ground, then builds the release pack and a signed decision record. Redact runs in the customer's country or the customer's own AWS account, and agencies start with a pilot on their own documents.
Alternatives to AccuFind Redact
Other products that do the same kind of work as AccuFind Redact, and what each one does.
- DDiligent EntitiesDiligent Entities manages subsidiaries, officers, ownership, filings and other corporate records for in house legal teams and corporate secretaries.
- EEthycaEthyca is a data governance and privacy engineering platform for enterprises. It is built on Fides, an open source privacy taxonomy and policy language that Ethyca wrote and maintains on GitHub.
- DDiligent BoardsDiligent Boards is a board portal where corporate secretaries and general counsel build board books, control who sees them, collect votes and signatures, and keep the meeting record.
- GGreenboardGreenboard is a compliance platform for firms regulated by the SEC and FINRA, built by Greenboard, Inc. of New York, founded in 2023.
- LLextreeLextree gives corporate counsel, corporate secretaries, compliance teams and law firms one record of entities, ownership, boards, signing authority, licenses and compliance obligations.
- PPrivIQ Data Privacy ManagementCompliance Technology Solutions B.V. sells PrivIQ Data Privacy Management to privacy officers, legal counsel and consultants.
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.
Smart Flags are the model driven core. Each one asks a single question about a single provision across every page or person in a case, using the officer's intake answers, and returns a strength of Strong, Possible or Weak with the likely paragraphs. Text recognition reads scanned pages, and person matching links each individual across the set, one worked example finding a person in 23 places across 49 documents. Around them sits a case workflow of intake questions, duplicate handling, decisions by category, sealing and release packs that would still run as a manual redaction tool without the models.
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.
Every flag is filed under a named provision, with examples drawn from sections 42, 45 and 47F of the Commonwealth FOI Act 1982 and section 552(b)(7)(D) of the United States FOIA on confidential sources. The likely paragraphs are highlighted on the page. The labeled release copy shows the ground for each redaction, and no redaction can be sealed without an accepted ground. No accuracy rate, missed exemption rate or test set accompanies the flags, and pilots end with measured results such as time saved and issues found, none of them released.
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 suggestion never redacts anything by itself, and officers apply, reject or dismiss every flag. Release decisions about a person are made by category, such as name and signature, contact details or role in the matter, and carry to every place that person appears. A ground that needs a public interest test asks for the factors first. Before sealing, Redact confirms every document is read, every suggestion answered, every new match decided, every redaction tied to an accepted reason and no document held, and sealing requires the officer to sign in again.
The record of decisions names who made each one and when, and officers can redact by hand or withhold whole pages or documents.
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.
No agency, pilot figure or customer account is tied to Redact. Its page shows worked examples only, such as a Strong flag on page 14 of a case under the confidential source exemption. The three stage pilot runs beside an agency's current process and ends in measured results such as time saved and issues found. The results of AccuFind's police and government trials belong to its Virtual Assistant.
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.
Prompts, responses and documents are never used to train AI models under AccuFind's terms, which limit the content license to running the service. Its privacy policy isolates each organization at the application, database and infrastructure layers, and the terms state that the companies making the AI models receive no client data. Audit logs stay seven years, and other data is deleted after a 90 day export window. Legal professional privilege is one of the exemptions Redact flags, under section 42 of the Commonwealth Act, and no separate commitment covers privileged material an agency loads into a case.
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.
Redact suggests and officers decide, and the terms call AI output a research aid that is not legal advice and that a professional should review before any decision. AccuFind's AI policy keeps final authority and control with legal professionals. Each deployment is configured to the jurisdiction's law and the agency's workflow, and grounds needing a public interest test require the officer's factors. Supported FOI regimes beyond the Commonwealth and American examples are not listed, and no rule sets the supervision a lawyer gives the release decision.
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.
AccuFind's AI policy, effective 20 February 2026, works to the nine AI principles of the Australia New Zealand Policing Advisory Agency, from transparency and explainability to human oversight, fairness and reliability, and keeps final authority with people. Company whitepapers explain how it evaluates grounded systems and takes AI into production. Redact adds a signed record of every decision with its maker and time, which an agency can check later.
The policy names no accountable owner and no testing schedule, and no check of how flags vary across document types, languages or agencies is described for Redact.
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.
AccuFind's terms give a 90 day export window after termination, followed by secure deletion confirmed in writing on request, with audit logs kept for seven years from creation and system logs for 30 to 90 days. AWS is the named subprocessor for infrastructure and model inference through Amazon Bedrock and Amazon SageMaker, new subprocessors come with at least 30 days' notice, and breaches are notified within 72 hours of confirmation.
Redacted text is removed from the file rather than covered, and the removal is verified before the pack is built. The officer downloads the pack and Redact sends nothing, and protective markings show on every screen.
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.
General liability is capped at the fees paid in the 12 months before a claim, rising to three times those fees for breaches of data security or confidentiality. Losses from relying on AI output without independent verification are excluded, and the terms warn that AI models may produce incorrect, incomplete or outdated information, with no accuracy warranty. South Australian law governs after 30 days of good faith negotiation, and an agency's own agreement with AccuFind takes precedence where one exists.
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.
Officers bring material in as PDFs, Office files, emails and mailboxes, zip and other archives, and scanned pages that text recognition converts. They take out a release copy, a labeled copy showing each ground, a schedule of documents, a marked up review copy and a signed decision record, downloaded by the officer rather than sent by the system. No connection to a records management system, an FOI case tracker or a document management system is named, and no API is offered.
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.
Redact runs either hosted by AccuFind in the customer's own country or deployed into the customer's own AWS account. AI processing happens in the AWS region where the product is hosted, the Sydney region for Australian customers or a United States region for Redact hosted there. The terms place model inference on Amazon Bedrock and Amazon SageMaker in the same region. Storage and processing share that region under AWS's data processing agreement, and Service Control Policies keep data from leaving Sydney for products hosted in Australia. Each organization is isolated at the application, database and infrastructure layers.
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.
No independent security certification or audit report covers Redact. The privacy policy names the controls behind it. Sign in runs through AWS Cognito with PKCE OAuth 2.0 and optional multifactor authentication by time based one time passwords, every database and store is encrypted with AES 256 through AWS Key Management Service, and access follows least privilege through AWS Identity and Access Management. Officers sign in to work a case and again before sealing, protective markings appear at the top and bottom of every screen, and decisions go into a signed record.
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.
Model inference runs on AWS through Amazon Bedrock and Amazon SageMaker in the hosting region. The terms state that the companies making the models receive no client data, and the privacy policy adds that Bedrock's terms bar training on customer inputs and outputs. The models behind Smart Flags, text recognition and person matching are not named, and the 30 day notice applies to new subprocessors rather than to a change of model.
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.
No price, plan or billing unit is listed for Redact, and AccuFind's fee terms cover only its Virtual Assistant and PolicyDNA, both under directly negotiated agreements. Agencies start with a pilot on their own documents in three stages. AccuFind maps the agency's process and law into the product while the agency's security review runs in parallel, staff then use Redact beside the current process, and the agency decides on measured results. Neither the pilot's length nor its cost is given.
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.
Government FOI, public records and release teams are the buyers, with examples under the Commonwealth FOI Act and the United States FOIA, where hosting is also offered. Each deployment is configured to a jurisdiction's law and an agency's workflow, and the exemptions shown span confidential sources, legal professional privilege, material given in confidence and personal privacy. AccuFind's legal collection reaches federal, state, territory and New Zealand law for its other products, but Redact's supported FOI regimes beyond the two Acts in its examples are not listed. Law firm or corporate use is not described.
5 public documents
The public pages on file for AccuFind Redact, with the recorded signals each one supports and the date it was last read. Open any of them and check the reading against the record.
-
accufind.ai/capabilities/redact5 signals
Primary Law Corpus Provenance, Good Law Verification, Refusal and Uncertainty Behavior and 2 more
Read Nov 20, 2026
-
accufind.ai/terms3 signals
Client Data in Training, Prompt and Output Retention, Outside Counsel Guideline Readiness
Read Nov 20, 2026
-
accufind.ai/privacy2 signals
Ethical Walls and Matter Segregation, Third Party Request and Subpoena Notice
Read Nov 20, 2026
-
accufind.ai/ai-policy1 signal
Bar Guidance Alignment
Read Nov 20, 2026
-
Fabricated Citation Record
Read Nov 20, 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?
The published terms prohibit training on customer content. Not a policy page, the agreement.
AccuFind's terms, effective 6 December 2026, state that prompts, responses and documents are not used to train AI models and limit the license over client content to running the service. The terms add that the companies making the models receive no client data, and the privacy policy that Amazon Bedrock's terms bar training on customer inputs and outputs.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
A specific retention period is published and the customer cannot change it.
Audit logs are kept for seven years from creation, and other organization data stays available for export for 90 days after termination before secure deletion. User account data stays for the life of the account plus 30 days after a deletion request, and system logs for 30 to 90 days. Redacted text is removed from release files rather than masked.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
The product maintains its own permission model, documented, requiring the firm to keep it aligned.
Each organization is isolated at the application, database and infrastructure layers, and work is organized case by case, with officers signing in to each case. No wall between teams or cases inside one agency is described.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Published terms or policy address disclosure to authorities or in response to legal process, and no commitment or reservation regarding customer notice is located anywhere. The vendor has told the customer that data can leave and has said nothing about whether the customer hears of it.
Disclosure is allowed where the law or valid legal process requires it, and nothing promises to tell the agency when an authority asks for its data. Notice is promised only for breaches and new subprocessors.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Coverage is described by jurisdiction with no identification of the underlying corpus.
Flags are tied to provisions of the FOI law each deployment is configured for, with examples from the Commonwealth FOI Act 1982 and the United States FOIA. No list of supported regimes or source for the statutory text is given. AccuFind's legislation and case law collection of 801,178 documents serves its Virtual Assistant, and no link between it and Redact's flags is described.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
No located public material addresses whether authority is checked for subsequent history.
Nothing tests whether a cited FOI provision has been amended or reinterpreted by a review body or court. Flags cite the provision itself, such as section 47F of the Commonwealth Act, with no link to decisions applying it.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
The product exposes a confidence or grounding score without an explicit abstention path.
Each flag carries a strength of Strong, Possible or Weak with the likely paragraphs, and officers decide every item. Duplicate, repeated and blank pages are set aside first, each with a reason. No path describes the system declining to flag a page it cannot read reliably.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
No court order, opinion or disciplinary record addressing fabricated or hallucinated legal citations produced by this product has been located as of the date shown. This is a statement about the public record on that one subject, not a finding about the product, and this signal is not a litigation history.
The AI Hallucination Cases database kept by Damien Charlotin has no entry naming AccuFind. Redact flags exemptions inside an agency's documents rather than drafting submissions.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Public materials refer to professional responsibility in general terms without naming guidance.
Final decisions stay with officers under AccuFind's AI policy, which follows the Australia New Zealand Policing Advisory Agency's AI principles and keeps final authority and control with legal professionals. No law society ruling, information commissioner guidance or other named ethics opinion is tied to Redact.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
The product does not touch a fee between a lawyer and a client. It operates before an engagement exists, or it is bought by a team that bills no client for the work. Savings claims aimed at the buyer’s own cost are recorded in the summary and do not make the row a savings claim, because no client bill is in the loop.
Government FOI and records teams buy Redact to release their own agency's documents, so no lawyer bills a client for the work. Agencies judge it on measured time saved and issues found during a pilot, with no figure tying that saving to the cost of a release.
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.
AWS is named as the subprocessor for infrastructure and model inference, through Amazon Bedrock and Amazon SageMaker, and new subprocessors come with at least 30 days' notice. The privacy policy names the security controls, AWS Cognito sign in, KMS encryption and least privilege access. No client facing disclosure pack is offered.
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.
Each sealed case yields a signed record of decisions naming who made each one and when, a labeled copy showing the ground for every redaction and a schedule of documents. The applicant receives the labeled copy and schedule, while the agency keeps the review copy and the signed record. The record does not state which model produced each flag.