State actors and financial fraudsters are exploiting the seam between security and HR. Census is the AI verification layer that closes it across the full hiring lifecycle, in language your auditors and your candidates can both follow.
Foreign IT-worker programs have placed operatives inside more than one hundred US enterprises, including security vendors and Fortune 500 firms. The schemes generate hundreds of millions of dollars per year in fraudulent salaries, and after detection, routinely escalate to data theft and extortion against the very employers being defrauded.
A single compromised security vendor is a multiplier across every customer it protects.
The capabilities to detect employment fraud already exist as point tools, owned by different teams, optimized for narrow windows. Individually they all return green. The joint pattern is the seam.
Education, employment dates, and certifications validated against source records.
Identity and criminal record check against the legal name provided.
Real-time liveness and deepfake-artifact detection during the live interview call.
Resume claims verify. Background record is clean against the provided identity. Liveness is real. And yet, the keystroke cadence on the take-home does not match the cadence captured during the interview, the LinkedIn timeline was assembled in a 6-week burst three months ago, and the candidate's GitHub commit hours sit in a UTC+8 distribution that nothing on the resume explains.
No single signal is enough. Census fuses four, each independently verifiable, each weighted, each surfaced as evidence on every decision.
Cross-checks claims against verifiable employment, education, and identity records, not just keyword overlap.
Detects deepfake artifacts and replay attacks across live and asynchronous interview footage.
Typing cadence, mouse dynamics, and tool fluency profile the operator behind the screen.
Continuously reconciles the resume timeline against public, social, and verifiable records over years.
Every score breaks down into the evidence that produced it. Borderline cases route automatically to human reviewers.
Catches the case point tools cannot: an operator substituted in after onboarding, or an account quietly transferred.
The moment of hire is no longer a hand-off out of risk visibility. Every score updates as new evidence arrives.
Census fits behind your existing stack. We orchestrate the deepfake detection, behavioral analytics, and OSINT layers you already pay for; we do not replace them. Continuous post-hire scoring catches the case point tools cannot: an operator substituted in after onboarding completes, or an account quietly transferred to a third party.
Adverse-action documentation is generated for you, with claim-by-claim detail specific enough to be challenged on its merits. Borderline scores are routed to human reviewers, not auto-resolved. And the platform produces one clean signal across the background-check, identity, and ATS tools your team already depends on.
Drop Census into your ATS, IDP, or onboarding flow with a few lines. Webhooks fire when evidence changes; no polling required.
import Census from "@census/sdk";
const census = new Census(process.env.CENSUS_API_KEY);
const { score, tier, evidence } = await census.evaluate({
candidate_id: "cand_01J5K9XQH2",
stage: "finalist",
});
if (tier === "borderline") {
await census.routeToReview(candidate_id);
}from census import Census census = Census(api_key=os.environ["CENSUS_API_KEY"]) result = census.evaluate( candidate_id="cand_01J5K9XQH2", stage="finalist", ) if result.tier == "borderline": census.route_to_review(result.candidate_id)
curl --request POST \
--url https://api.census.com/v1/evaluations \
--header "Authorization: Bearer $CENSUS_API_KEY" \
--data '{
"candidate_id": "cand_01J5K9XQH2",
"stage": "finalist"
}' If you are a security vendor, a Fortune 500 employer, or a policy body that takes employment-vector intrusion seriously, we want to talk. Development partners get hands-on access, direct influence on the roadmap, and a seat at the table on the cross-industry signal-sharing schema we are convening.
Covers the threat model, the platform's design principles, the scoring methodology, and our proposal for cross-industry hiring-risk signal sharing.
State actors and financial fraudsters are exploiting hiring as a cyberattack surface. The North Korean IT-worker program alone has placed operatives inside more than one hundred US enterprises, including security vendors and Fortune 500 firms, generating hundreds of millions of dollars per year in fraudulent salaries and, after detection, escalating to data extortion against the very employers being defrauded. The capabilities required to detect this threat exist today as point tools (resume validation, background checks, document authenticity, deepfake detection, behavioral biometrics) but they do not communicate with one another, and no system reasons over them coherently across the full hiring lifecycle. Census Networks is an AI-first platform that fills that gap. Multiple specialized models, organized into four pillars, produce independent judgments that a fault-tolerant consensus engine fuses into a single, explainable risk score that travels with a candidate from application through continuous post-hire monitoring. This paper sets out the threat model, the platform's design principles, the scoring methodology and compliance posture, a concrete proposal for cross-industry hiring-risk signal sharing, and the deployment and integration patterns that make the platform real.