AccessPoint vs. Veritone Redact

The short answer

Veritone Redact is a mature, vendor-hosted AI redaction application for audio, video and image evidence, used by a large base of US public-safety agencies and resold as an integration by FOI request platforms such as JustFOIA and OPEXUS FOIAXpress. If your office already licenses one of those platforms and is content to send footage out to Veritone's cloud for processing, it is a credible way to add automated head and plate masking. AccessPoint's difference is where the work happens and what survives it: the recording never leaves your tenant, every mark carries the exemption it relies on, and the burned release, its hash manifest and a time-based redaction log stay in the same request file as the paper records — with no separate seat licence.

What is Veritone Redact (Veritone, Inc.)?

Veritone Redact is Veritone's AI-assisted redaction application for public-safety and legal evidence. Veritone's own documentation describes it detecting human heads, licence plates and mobile data terminals, letting users define other sensitive objects in a scene, and then redacting that material from audio, video and image files; its product page adds time-synced transcripts, voice masking and multilingual transcription on the audio side, and describes detailed logs, chain-of-custody records and downloadable audit trails. In July 2026 Veritone added document redaction, positioning Redact as a single hub for image, audio, video and document work.

The product is sold to agencies directly — Veritone says it is trusted by more than 300 federal, state and local government agencies, and announced a multi-year contract with the California Highway Patrol in June 2026 — and indirectly through FOI request platforms. OPEXUS markets a FOIAXpress partnership in which audio and video files are sent to Veritone Redact with one click, and in August 2026 Veritone announced an integration with MCCi's JustFOIA for state, local and education agencies. Veritone also offers a Redaction Managed Service for offices that would rather outsource the work.

That reseller position is the comparison point. When the request platform hands footage to a separate redaction service, the recording is processed in a second vendor's cloud and the record of what was withheld — and why — has to travel back across the integration. AccessPoint's position is that the mark, its exemption, its approval, the burned release and the redaction log should be one record in one system, and that the system should run inside the tenant the office already controls.

Veritone Redact fact sheet

Product: Veritone Redact — AI redaction for audio, video, image and (since July 2026) document files, plus an optional Redaction Managed Service
Vendor: Veritone, Inc.
Form factor: Vendor-hosted web application, part of Veritone's public-sector suite; also embedded in FOI request platforms as an integration
Processing location: Veritone's cloud; its public-sector materials describe CJIS-compliant cloud hosting. A specific region, government-cloud option or customer-tenant deployment is not described in its published documentation
Pricing model: Not described in its published documentation; sold directly and through FOI platform integrations
Notable customers and integrations: More than 300 government agencies (vendor claim); California Highway Patrol (multi-year contract, June 2026); US Department of Veterans Affairs (representative customer in OPEXUS materials); integrations with MCCi JustFOIA (August 2026) and OPEXUS FOIAXpress

Sources

Veritone Redact and AccessPoint, side by side

	Veritone Redact	AccessPoint
Where the footage is processed	Uploaded to Veritone's cloud — directly or through the request platform's integration. Veritone describes CJIS-compliant cloud hosting; region and sovereignty options are not described in its published documentation	Inside your own Microsoft 365 and Azure tenant. Face detection runs on your App Service with no outside call; on-screen text detection and transcription use Azure AI Document Intelligence and Azure AI Speech deployed in your tenant. The original recording is never modified
Detection: faces, plates, on-screen text	Automatic detection of human heads, licence plates and mobile data terminals, with user-defined objects for anything else; Veritone's CHP announcement also cites faces, medical information and driver's licence numbers	In-tenant face detection with stabilised tracking (one proposed tracked mark per person) and on-screen text via frame OCR against your detection patterns — a licence plate, a name badge, a case number on a monitor — each proposal carrying a rule citation. No dedicated plate, head or body detector; anything else is drawn once and tracked automatically
Audio: mute, bleep, voice distortion	Automatic detection of sensitive speech and PII; product page describes voice masking and multilingual transcription. Which obscuring effects are available per span is not itemised in its published documentation	Mute, bleep and voice distortion (pitch-shifted so every word stays audible) can be combined on one recording, each span with its own exemption — the severance Ontario's IPC has ordered in place of removal
Transcript-anchored marks	Time-synced transcripts are part of the product; the Escondido PD case study describes time-correlated transcription of sensitive content	Time-coded, speaker-labelled transcript; Find & redact, detection rules and the AI pass propose Mute marks over the spoken span for you to accept or dismiss. Needs Azure AI Speech deployed in your tenant
Link to the request and the exemption cited	The FOI-platform integrations describe a one-click round trip of the file. Whether a per-span legal justification travels back into the request file's own withholding record is not described in the published integration materials	Every mark — region, audio span or whole-record withhold — carries the exemption, alternates, comments, flags and approval like a page redaction, the in-frame caption can show it, and the request's redaction log lists each span with its effect and exemption
Audit log and release record	Detailed logs, chain-of-custody records and downloadable audit trails, held in Veritone's application	Hash-chained, append-only audit ledger in the request file; each release carries a MANIFEST.sha256 of every entry, a Redaction Index and a "How this release was prepared" statement that counts manual, rule-proposed, model-proposed and propagated marks
Formats and the burned release	Audio, video and image files are supported; the redacted file is returned to the agency or the request platform. Accepted formats, size limits and the shape of the released file are not described in its published documentation	Uploads up to 2 GB by default in resumable blocks; the release renders MP4 / MP3 with marks burned in and a record-label caption, packaged in the same ZIP as the PDFs. Per-audience release variants are stored as exclusions, so a new mark applies to every cut by default
Integration with request platforms	Integrations announced with MCCi JustFOIA (August 2026) and OPEXUS FOIAXpress; the request lives in one system and the redaction in another	AccessPoint is the request platform: recordings are documents on the request, added from the Documents tab or Teams meeting recordings, and the release decision feeds the Contents page, the redaction log and the statistics
Pricing model	Not described in its published documentation; typically an additional licence alongside the request platform	Flat annual subscription by organization size, published on the site and billed through the Microsoft commercial marketplace — no per-user, per-seat or per-minute fees, and video and audio redaction is included

Competitor capabilities vary by product edition and configuration; this table reflects each product's public positioning as of 2026-09-04.

Where Veritone Redact fits

A fair comparison cuts both ways — Veritone Redact is a reasonable choice when:

Your office already runs JustFOIA or FOIAXpress, the Veritone integration is licensed, and staff are comfortable with footage being processed in Veritone's cloud.
You want automated masking of heads and licence plates across a high volume of body-worn footage today, and a US-hosted, CJIS-aligned vendor cloud satisfies your security review.
You would rather outsource the work entirely — Veritone's Redaction Managed Service is an option AccessPoint does not offer.
Your redaction workload is mostly evidence preparation for prosecution rather than access-to-information releases, so a link to the request file and its exemption catalogue matters less.

Where AccessPoint differs as a Veritone Redact alternative

The recording never leaves your tenant Detection, transcription, review and the burned release all run inside your own Microsoft 365 and Azure tenant — no second vendor's cloud in the chain of custody, and your existing security stack governs every step.

The exemption rides on the mark Each region, audio span or whole-record withhold carries the section it relies on, alternates and an approval, and the caption burned into the frame can show it — the citation is made at the place in the record where the deletion is made, not applied to the whole file afterwards.

One record, not a round trip The recording is a document on the request. Its marks, the release ZIP with its hash manifest, the Redaction Index and the time-based redaction log stay in the same case file as the paper records — nothing has to be re-keyed from another system's report.

No per-seat add-on Video and audio redaction is part of the platform, priced flat by organization size and published on the site. AccessPoint is listed on the Microsoft commercial marketplace, reviewed and tested by Microsoft, with a 30-day trial that runs in your own tenant.

Migrating from Veritone Redact to AccessPoint

There is no case history to move out of a redaction tool. Releases you have already made through Veritone Redact stay where they are; the practical test is to run the next request with recordings through AccessPoint beside it and compare what comes out.

Keep prior releases and their Veritone audit trails where they are — nothing needs to be re-imported for the record to stand.
Deploy Azure AI Speech and the media processor in your tenant (both are deployment parameters), then take the next request that includes body-worn, CCTV or interview audio and work it in AccessPoint in parallel.
Run face and on-screen text detection and a transcript-based Find & redact, then accept or dismiss the proposals with the exemption on each mark.
Compare the two releases side by side: the burned file, the hash manifest, the redaction log's per-span effect and exemption, and the reviewer hours against the media review estimate.
Retire the Veritone licence at renewal once the office is satisfied — the request platform integration can simply go unused until then.

If your request platform is also being replaced, the request history moves through AccessPoint's Excel-based data import; recordings from past releases can be staged in your own Azure storage with hash verification.

Plan your migration → · Data Import & Export guide · Migration & onboarding services

Questions people ask about Veritone Redact and AccessPoint

What is Veritone Redact?

Veritone Redact is Veritone's AI-assisted application for redacting sensitive material from audio, video, image and — since July 2026 — document evidence. Veritone's documentation describes it detecting human heads, licence plates and mobile data terminals, letting users define other objects to mask, producing time-synced transcripts with voice masking, and keeping detailed logs, chain-of-custody records and downloadable audit trails. It is sold to agencies directly, through a managed redaction service, and as an integration inside FOI request platforms such as JustFOIA and FOIAXpress.

Is there a Veritone Redact alternative that keeps footage in our own tenant?

Yes. AccessPoint runs video and audio redaction inside your own Microsoft 365 and Azure tenant: the face detector runs on your App Service without calling an outside service, on-screen text detection and transcription use Azure services deployed in your tenant, and the original recording is never modified. There is no second vendor cloud in the chain of custody, and there is no separate seat licence — video and audio redaction is part of the platform.

Does AccessPoint detect faces and licence plates automatically like Veritone Redact?

AccessPoint detects faces in-tenant and proposes one tracked mark per person for you to accept or dismiss, and reads on-screen text with frame OCR so a licence plate, a name badge or a case number on a monitor becomes a proposal carrying a rule citation. It does not ship a dedicated licence-plate, head or body detector; a subject seen from behind is drawn once and then tracked automatically in both directions. Every proposal is a suggestion until a reviewer accepts it — nothing is applied silently.

Our request platform already integrates with Veritone. Why would we change?

The integration sends the recording out and brings a redacted file back. What it does not do, in the published integration materials, is carry a per-span legal justification into the request file's own withholding record. In AccessPoint the mark, its exemption and alternates, its approval, the burned caption, the release package and the time-based redaction log are one record that never crosses a vendor boundary — which is what a regulator or a requester challenging the severance will ask to see.

How does AccessPoint handle voice distortion compared with Veritone Redact?

Veritone's product page describes voice masking as part of its audio capability. AccessPoint offers Mute, Bleep and Distort voice as effects on any audio span: Distort pitch-shifts the span so the speaker cannot be identified while every word stays audible, which is the severance the Ontario IPC has ordered in place of removal and one of the obscuring methods Washington's public-records fee statute lists. The three effects can be combined on the same recording, and the redaction log records the effect and exemption for each span.

Competitor information on this page is drawn from public sources — vendor websites, government reports, and press coverage — and was last reviewed in 2026-09-04. Products evolve, and a comparison is never the whole story: capabilities vary by edition and configuration. Spot something out of date? Tell us and we'll correct it.