FOIA & ATIP Requests | Webinar Recording

Summary

In this session, Duncan Fraser from Noticia looked at how to handle Freedom of Information (FOI/FOIA) and Access to Information & Privacy (ATIP) requests in Nuix Neo Discover. He framed public disclosure work as effectively the same job as civil eDiscovery - collect documents, protect what should not be produced, do it within a deadline, and avoid failure - and made the case for bringing the two functions closer together given how similar the skills and technology are.

Attendees saw a live walkthrough of an ATIP review in a demo case: preparing and winnowing a document set, setting up exemption codes and a security override binder, tracking work with coding clarifications, redaction terms and consultation status, applying redactions with Find and Redact, producing burned-in image PDFs, and using AI to identify and summarize content that may need to be redacted.

Key Areas Covered

Why FOI/ATIP and eDiscovery Belong Together:

• FOI/ATIP and civil disclosure sit on the same spectrum of legal obligations to collect, review, and produce information under a statute or rules, with the ability to redact private, privileged, and sensitive content, all within a defined deadline.

• Civil litigation disclosure has historically attracted more funding and better technology because the perceived risk is higher. The skill sets and tooling overlap so heavily that it makes sense to relate or integrate the two functions in public organizations.

Focus on Document Review, Not Case Management:

• Requests about intake portals, fees, and requester communication are common, but Nuix is built for document review and review management - that is where the time, money, and efficiency gains actually are.

• Case management often gets the budget attention because it is what management reports on, but there is little efficiency to gain there. Keep the business case focused on the review side.

Building the Business Case:

• Benchmark first - know what you do today, what it costs, and how many inadvertent disclosures are caught in QC versus first pass. You will need this to convince management to fund new tooling.

• Then evaluate alternatives, optimize your process to take advantage of the technology, and integrate it into the organization. Streamline the inputs and outputs, since review is one part of a larger process. Document your workflows and secure and train the right talent.

Preparing and Winnowing the Document Set:

• Public disclosure is generally produced as flat, imaged PDFs (not TIFFs). As administrator, prepare the set fully: deduplicate at the family level, decide on email threading, and ensure everything is imaged or has a plan for files that do not image well (audio, video, Excel).

• Winnow 13,000 documents down to the responsive set using family deduplication, email threading, and date, custodian, and domain filters. Place the responsive documents in a binder so you always know what is in scope.

• Ensure extracted text (OCR) and imaging are complete on the working set - extracted text is what the AI tools rely on later.

Case Setup, Security, and Review Fields:

• Use a security override binder so a review group can only see the in-scope documents - no matter what they search, they are locked out of everything else. The case admin controls what goes into the binder.

• Set up exemption codes as coloured highlights (not black redactions) so work is easier to track, and only configure the levels your team actually applies. Keep one decision field that covers every document (release in full, consult, redact, not relevant) plus a QC/complete field.

Tracking Workflow: Clarifications, Terms, and Consultations:

• Use a coding clarification field to flag documents where a reviewer is unsure - a manager can pull up all open clarifications, answer them, and the answers become a shared record for the whole team.

• As recurring names or terms appear (for example a witness whose name must be redacted throughout), capture them in a redaction terms field to build a running list for the set.

• Track consultation status (internal and third-party) with notes, so you know which documents still need sign-off before production.

Applying Redactions and Producing:

• Find and Redact (one document at a time) supports word/phrase, regular expression, search term family, and AI entity modes. It applies presumptive redactions, but because it is a search you still need to QC for spelling, order, and OCR variants. Pin the colours you use most to keep the list manageable.

• Produce with a saved search (release in full or redaction applied, and QC complete) via custom export as a burned-in image PDF. Omit native and text files where redactions exist, turn off OCR, and add a header. Proof mode gives a see-through version for supervisor review.

• Use the related/similar documents dashboard as a QC check to catch near-duplicates you may have coded differently, and download the highlight list to confirm which annotations sit on which documents.

AI-Assisted Redaction:

• The AI runs on a private, controlled data centre with no learning on your data. You can have it generate and refine a prompt (for example, content to redact under the Access to Information Act, with section numbers in the output) by talking to it conversationally.

• Scan a single document to see flagged content with justification and quotes, or run a bulk scan across the production set that writes searchable, viewable results to an AI summary field.

• Once agreed, the flagged text can be prepared into a search term family for auto-redaction. The point is not to replace the reviewer - AI enriches and speeds the work, and the human still makes the decision.

Follow-Up Notes from the Session

A few items were flagged as out of scope for this session: Excel handling and the native Excel redaction tool (a per-document Nuix add-on, not included in hosting) will be covered separately, as will regular expressions and collection strategy - contact us if you would like a walkthrough of any of these. This was the last of the regularly scheduled monthly sessions; the series resumes in September with a new schedule based on client feedback, so send us any topic suggestions. If you would like to know more about the AI tool, let us know and we will share dedicated recordings and webinars.