Federal Data Exchange — Privacy & Transparency Platform
A FOIA-compliant interface for individual-level government records.
Submit natural-language queries — our AI matches the right dataset and automatically
applies two-tier redaction before returning results.
Two-tier FOIA redaction (5 U.S.C. § 552):
Classified, statutory, and operationally sensitive fields are
blind-redacted with [b(Ex.N)] markers citing the
applicable exemption — e.g. SSNs become [b(Ex.3)],
clearance levels become [b(Ex.1)].
Personal privacy fields (names, addresses, dates of birth, phone numbers) are
smart-redacted — replaced with realistic equivalents so record
structure is preserved: “Marcus Thompson” becomes “David Ellison”,
“412-67-8234” becomes [b(Ex.3)].
All non-exempt fields — case numbers, dates, titles, pay grades, dispositions —
are released in full per the FOIA segregability requirement.
Powered by Acme Redactors
The redaction engine behind Poseidon is
Acme Redactors — an AI system purpose-built for FOIA compliance
processing. Acme Redactors deploys a cloaking device: a layered identity-masking
system that camouflages and protects the identities of both public servants and US citizens
appearing in federal records. Rather than simply blacking out sensitive fields, the cloaking
device substitutes realistic cover identities — different names, dates of birth, and
contact details — ensuring that no individual can be traced through released documents
while preserving the structural integrity of every record for legitimate oversight and research.
Acme Redactors ingests raw federal records, classifies each field against
the applicable exemption schedule, and applies the two-tier scheme automatically:
blind [b(Ex.N)] markers for classified or statutorily protected values,
and realistic cover identities for personal privacy fields. It preserves all
segregable non-exempt content per § 552(b), returning a release-ready JSON document
alongside the original for auditing.
Features
AI Dataset Matching
Describe records in plain English — AI selects the right dataset.
Cloaking Device
PII replaced with realistic substitutes, not blacked out.
Cloud Infrastructure
Digital Ocean Spaces with S3-compatible object storage.
Open Access
Try the demo freely; register for API key access.
Try It
Demo Queries
Choose a dataset to load a sample query and run it.
Each dataset demonstrates a distinct FOIA exemption profile.
⚠ All datasets below are AI-generated synthetic records created for demonstration purposes only.
They are not real government data, do not represent real individuals, and are not derived from any actual public records release.
Paste a Record
Paste any text or JSON. Aggressive cloaking is always applied:
names, nicknames & aliases, dates of birth, addresses, phone numbers, email addresses,
locations, event dates & times, employers, organizations, schools, military units,
family members & associates, vehicle descriptions, physical descriptions, immigration status,
language & nationality details, financial specifics, and any other field that could logically
be cross-referenced to re-identify the individual.
Classified and statutory values receive blind [b(Ex.N)] markers.
Sign Up / Log In
Create an account for API key access and saved history.
Raw Record (unredacted)
FOIA-Processed Release
Redaction Key — 5 U.S.C. § 552
[b(Ex.N)]FOIA blind redaction — value suppressed entirely.
Ex.1: classified info (markings, clearance levels, covert facilities).
Ex.3: statutorily protected (SSNs, program identifiers).
Ex.7(F): information that could endanger life or safety.
David EllisonStandard cloaking — realistic substitute injected.
Ex.6: names, addresses, DOB, phone, email replaced with different plausible equivalents.
Ex.7(C): third-party names in law enforcement records.
PortlandAggressive cloaking — additional fields substituted to prevent re-identification.
Locations, event dates/times, nicknames, employers, organizations, relationships,
physical descriptions, and any detail cross-referenceable against public records.