---
title: "The insurance fraud referral package was never the bottleneck: 51,419 reports in, 243 cases opened"
description: "New York's fraud bureau received 51,419 reports of suspected fraud in 2025 and opened 243 cases. An insurance fraud referral package is an input to a decision, not an outcome."
date: "2026-10-12"
lastModified: "2026-11-16"
author: "Pankaj Dhariwal"
tags: ["Vendor Landscape"]
canonical: "https://gethesperai.com/blog/siu-fraud-referral-package-automation/"
---

# The insurance fraud referral package was never the bottleneck: 51,419 reports in, 243 cases opened

> **TL;DR** Shift Technology shipped a Lead Package Agent on October 1, 2026 that assembles an insurance fraud referral package out of an alert, and it is well-scoped work. What it does not touch is conversion: New York's fraud bureau received 51,419 reports of suspected fraud in 2025 and opened 243 cases, and insurers reported $40.3 million of recoveries against $3.5 billion of claimed savings.
>
> - 51,419 insurance fraud reports in, 243 cases opened, 2025
> - Automated tools: 45% of SIU referrals, 29% of accepted ones
> - $40.3 million recovered against $3.5 billion of claimed savings

- **51,419** - Reports of suspected fraud received by New York's Insurance Frauds Bureau in 2025 (243 cases were opened for investigation in the same year. NY DFS annual report, updated April 15, 2026)
- **45% / 29%** - Automated tools' share of SIU referrals, and their share of accepted referrals (Acceptance ratio 36%, against 70% for adjuster referrals. Coalition Against Insurance Fraud and Aon, 2024 study, 35 P&C carriers)
- **60 days** - Statutory window to pay or deny a Florida property claim, referral or no referral (Fla. Stat. 627.70131(7)(a). California separately allows 60 days from reasonable belief to file the referral, Cal. Ins. Code 1872.4)
- **$40.3M** - Recoveries from SIU investigations New York insurers reported for 2024 (Against $3.5 billion in SIU-attributed savings reported by the same insurers for the same year. Derived: 1.2%)

The insurance fraud referral package has just become a product category. On October 1, 2026, Shift Technology [introduced its Lead Package Agent](https://www.shift-technology.com/resources/reports-and-insights/introducing-shifts-lead-package-agent), which takes an alert something else raised and assembles a referral-ready case package for an investigative team. It is well-scoped, genuinely useful work.

The question a claims or SIU leader has to answer is a different one: what happens to the claim after the package exists. A referral documents suspicion. It does not take the coverage position, move the reserve, pay, deny or recover a dollar. For the handler-assist side of Shift's catalogue, we covered [Shift Claims against autonomous investigation](/blog/shift-claims-agentic-ai-vs-hesper/) separately.

## Where insurance fraud referrals go, and how few come back

**What happens to an insurance fraud referral package after it is filed?**

An insurance fraud referral package is the documented case file a carrier sends to its Special Investigation Unit and, where the facts support it, to a state fraud bureau. New York's bureau received 51,419 of them in 2025 and opened 243 cases for investigation, so roughly one report in 212 became a state investigation.

New York reports both ends of the funnel for the same year. The Department of Financial Services' annual report on its Insurance Frauds Bureau, in the [current edition, updated April 15, 2026](https://www.dfs.ny.gov/reports-and-publications/dfs-annual-reports/cpfed-annualrep-2025), states: "In 2025, IFB received 51,419 reports of suspected fraud" and "In 2025, IFB opened 243 cases for investigation, affected 169 arrests, and referred 95 cases to prosecutorial agencies for prosecution."

That is 0.47% of reports, about one in 212, becoming a state investigation, and 0.18%, about one in 541, reaching a prosecutor. Both ratios are derived from the bureau's own figures. A referral is not a charge, and the bureau is not failing at anything. It is behaving like an intelligence intake.

Drawn to a single scale, the 2025 funnel is hard to look at.

| New York Insurance Frauds Bureau, 2025: reports received to prosecutor referrals, drawn to real scale | Value | Share |
| --- | --- | --- |
| Reports of suspected fraud received | 51,419 | 100% |
| Cases opened for investigation | 243 | 0.47% |
| Arrests affected by the bureau | 169 | 0.33% |
| Cases referred to prosecutors | 95 | 0.18% |

What has changed is the ratio between supply and capacity. In 2020 the bureau received 30,113 reports and opened 324 cases. By 2025 reports received had grown 70.8% and cases opened had fallen 25.0%, both derived. Every additional referral filed since 2020 has landed in a system whose throughput shrank by a quarter.

## What an insurance fraud referral package must contain

**What must an insurance fraud referral package contain?**

In California, regulation prescribes the fields: fraud type designation, reporting party information, alleged victim details, policy and claim specifics, loss amounts, a complete factual synopsis and party identifiers, under 10 CCR 2698.37 and 2698.38. California Insurance Code 1872.4 then gives the insurer 60 days from its determination of reasonable belief to file it with the Fraud Division.

A referral package in a P&C claim is a regulated artefact, not a memo whose shape is up to the carrier. California's rules are the most prescriptive of any state. Under [10 CCR 2698.30](https://insurancefraud.org/regulations/california-siu-requirement-state-insurance-code-section-1875-20-california-code-of-regulations-subchapter-9-insurance-fraud-section-2698-30-et-seq/), a "reasonable belief" is "a level of belief that an act of insurance fraud may have or might be occurring for which there is an objective justification based on articulable fact(s) and rational inferences therefrom".

The deadline is statutory: [California Insurance Code 1872.4](https://codes.findlaw.com/ca/insurance-code/ins-sect-1872-4/) requires notice to the Fraud Division "within 60 days after that determination by the insurer". We walk the whole regime in [our guide to 10 CCR 2698 SIU compliance](/blog/california-10-ccr-2698-siu-compliance/).

The Department's [SIU Compliance Unit](https://www.insurance.ca.gov/0300-fraud/0100-fraud-division-overview/12-siu/) says its office evaluates "the accuracy, completeness, and timeliness of the report filed by approximately 1,100 insurers each year". 10 CCR 2698.40 also requires an SIU annual report within 90 days of the Department's notification, and referral statistics are a filed number in it.

**Figure 1 (The regulated artefact) - The fields are prescribed, and the clock starts at reasonable belief.** A referral package is not a free-form memo. California sets the trigger, the contents, the form and the deadline.

**The regulated path of one referral**

1. Reasonable belief forms: objective justification from articulable facts
2. The package is assembled with the fields the regulation prescribes
3. Filed with the Fraud Division within 60 days of that determination
4. The state decides whether to open a case of its own

**The California referral, by citation**

- Trigger: reasonable belief of insurance fraud: 10 CCR 2698.30
- Prescribed contents of the referral: 10 CCR 2698.38
- Notice to the state Fraud Division: Within 60 days
- Form the referral is filed on: eFD-1, FD-1, SFC
- Insurers whose reports CDI evaluates each year: About 1,100
- States, plus DC, running an insurance fraud bureau: 42 plus DC

*California is the most prescriptive regime. The last row is the national footprint, counted by the NAIC.*

**The clock:** The 60 days run from the carrier's own determination of reasonable belief, not from the date of loss.

Sources: 10 CCR 2698.30, 2698.38 and 2698.40 via the Coalition Against Insurance Fraud's California SIU regulation page; Cal. Ins. Code 1872.4 for the 60 days after the insurer's determination; California Department of Insurance SIU Compliance Unit page for the eFD-1, FD-1 and SFC forms and the approximately 1,100 insurers whose reports it evaluates each year; NAIC insurance fraud topic page, updated June 23, 2026, for 42 states and the District of Columbia.

New York regulates the apparatus rather than the document. Insurance Law section 409 requires qualifying insurers to file a Fraud Prevention Plan that "must provide for a full-time Special Investigations Unit", with an annual report on the plan due by March 15. The detail is in [our breakdown of the New York Fraud Prevention Act SIU requirements](/blog/new-york-fraud-prevention-act-siu/).

Nationally, the [National Association of Insurance Commissioners](https://content.naic.org/insurance-topics/insurance-fraud), on a topic page last updated June 23, 2026, records that "42 states and the District of Columbia operate insurance fraud bureaus that investigate claims of illegal insurance activities".

## What Shift's Lead Package Agent does, in Shift's words

**What does Shift Technology's Lead Package Agent do?**

Shift Technology's Lead Package Agent, introduced on October 1, 2026, starts from an alert another system already raised and assembles a referral-ready case package for an investigative team to decide on. Shift says it pulls together claims data, public records, credentialing checks and adverse media, and that every finding traces back to its source.

Shift writes that the agent "turns a single alert into an expert-ready package in hours instead of weeks" and that "every finding traces back to its source, so the investigator stays fully in control of the decision and the agent simply does the assembling".

Shift is also right that this puts scheme knowledge and data access in front of a six-person SIU that never had them. And the page is disciplined: no claim of autonomy, no claim to resolve or close a claim, and no number attached to "hours".

Shift is not a single-agent company either. Its Solutions navigation describes agent lines across most of the claim, in its own words.

| Shift solution line | Shift's own description of it |
| --- | --- |
| Fraud & Risk | "Agents to assess for fraud, automatically assign cases, guide investigations, and autonomously adapt fraud strategies" |
| Coverage & Liability | "Agents to assess policy coverage and liability, potential for agreement, and evidence strength" |
| Subrogation | "Agents to assess subrogation opportunity, create initial demand package, review third party insurer response" |
| Injury | "Agents to assess applicability of medical billing to a claim, estimate medical duration, and develop treatment plan" |
| Process | "Agents to accelerate the claim handling process, mitigate loss, and identify customer satisfaction concerns early" |
| Insurance Data Network | "Agentic intelligence layer that assesses cross-carrier claim history, synthesizes key insights, and recommends relevant next actions" |

The clearest statement of this one agent's scope is Shift's own. Writing about a social adult day care payment-integrity example on a [companion page dated October 6, 2026](https://www.shift-technology.com/resources/reports-and-insights/taking-on-social-adult-day-care-investigations-with-shifts-lead-package-agent), Shift says: "That transparency keeps the investigator in charge of the decision: the agent handles the assembly, and a senior reviewer still decides what gets referred and how it moves forward." The agent's job ends at that decision.

## More packages does not mean more cases

**Do automated fraud referrals get accepted as often as adjuster referrals?**

No. The Coalition Against Insurance Fraud and Aon's 2024 Insurer SIU Benchmarking Study of 35 P&C carriers found automated fraud detection tools generate 45% of all SIU referrals but only 29% of accepted ones. Their acceptance ratio is 36%, against 70% for adjuster referrals, so they convert at roughly half the rate.

Between the package and the case sits an acceptance gate. The [2024 Insurer SIU Benchmarking Study](https://insurancefraud.org/wp-content/uploads/2024-Insurer-SIU-Benchmarking-Study.pdf), run by Aon with the Coalition Against Insurance Fraud across 35 participating P&C carriers, reports that automated referrals rose to 45% of all referrals to SIU, with adjuster referrals dropping to 49%. Of accepted referrals, adjusters account for 60%, automated tools 29%, and all other sources 10%.

The acceptance ratios explain the gap. The study puts adjuster referral acceptance at 70%, down from 76% in 2022, and the automated-tool ratio at 36%, up from 32%. So 45% of referral volume at a 36% acceptance ratio contributes 16.2% of all referrals as automated and accepted, while 49% at 70% contributes 34.3%. Automated referrals are still accepted at roughly half the adjuster rate.

**Figure 2 (Referral acceptance, 2024 study) - Automated tools send 45% of referrals and win 29% of the acceptances.** A 36% acceptance ratio on 45% of supply lands on the 29% share of accepted referrals the study reports independently.

45% times 36% equals 16.2%

- Automated share of SIU referrals: 45% (Up 6 points since the 2022 study)
- Acceptance ratio, automated tools: 36% (Up from 32% in the 2022 study)
- Share of all referrals, accepted: 16.2% (Adjusters: 49% at 70% gives 34.3%)

**Referral acceptance ratio**

- Adjuster referrals: 70% - Down from 76% in the 2022 study
- Automated fraud detection tools: **36%** - Up from 32% in the 2022 study

**Net effect:** Automated referrals convert into accepted referrals at roughly half the adjuster rate, derived two ways from one study.

Source: Coalition Against Insurance Fraud and Aon, 2024 Insurer SIU Benchmarking Study, September 2024, 35 participating P&C carriers. Referral share and acceptance ratio are separate measures in the study; the product of the two is derived here. Accepted referrals by source are reported as adjusters 60%, automated fraud detection tools 29%, all other sources 10%.

That does not mean automated referrals are worse. A flag and a case are different objects, and the unit decides which flags become cases on a fixed budget of investigator attention. The same study measures that budget: carriers averaged 0.67 FTE per $100 million of gross premiums written against 0.9 two years earlier.

Practitioners have said as much for years.

> One of the biggest concerns that SIUs have is they are going to have more cases than they can handle. So it's kind of a chicken or the egg thing.
>
> - Dennis Jay, the retired executive director of the Coalition Against Insurance Fraud, speaking to Insurance Journal, August 25, 2020

Referral volume and quality do not translate into cases, findings and dollars unless something changes at the receiving end. That is the same bottleneck we measured in [why most flagged insurance claims are never fully investigated](/blog/why-flagged-insurance-claims-never-investigated/).

## A referral does not stop the claim clock

**Does referring a claim to SIU pause the claim's statutory deadlines?**

No. The referral clock and the claim clock run in parallel, not in series. Florida requires an insurer to acknowledge a claim communication within 7 calendar days, begin a reasonably necessary investigation within 7 days of receiving proof-of-loss statements, and pay or deny a property claim within 60 days of notice.

This is the whole argument in one statute. [Florida Statute 627.70131](https://www.flsenate.gov/Laws/Statutes/2024/627.70131), subsection (7)(a), reads: "Within 60 days after an insurer receives notice of an initial, reopened, or supplemental property insurance claim from a policyholder, the insurer shall pay or deny such claim or a portion of the claim unless the failure to pay is caused by factors beyond the control of the insurer."

California gives the carrier 60 days from its own determination of reasonable belief to notify the Fraud Division. Florida's clock started at first notice. The referral clock starts when suspicion crystallises, which is usually well after. Filing the referral consumes time on the disposition clock and buys none.

An adjuster whose coverage decision is due in 60 days gains nothing from a faster referral unless the same evidence pass also produces the coverage position, the reserve note and the documented reasoning behind them. We treated the reporting-deadline side of this in [how AI agents compress SIU reporting timelines](/blog/ai-agents-siu-reporting-timeline-compression/).

The destination of an SIU investigation is a decision about payment, not a referral.

> The best SIU investigation seeks a logical and reasonable explanation for payment of the claim.
>
> - Matthew J. Smith, Esq., Executive Director, Coalition Against Insurance Fraud, writing in CLM Magazine, March 8, 2024

The artefact that supports that decision is not the referral package. It is a complete file: the policy language, the material misrepresentation, the payment ledger, the evidence of the deliberate act, and the documented reasoning that connects them.

That file is what [our fraud investigation module](/platform/fraud-investigation/) is built to produce. It is also the thing a manual unit spends most of its 14+ days per case assembling, a Hesper internal benchmark that also frames [the comparison against manual investigation](/compare/hesper-vs-manual-investigation/).

*Figure: 51,419 reports in. 243 cases opened. Assembly was never the bottleneck.*

## A fraud finding is not a recovery

**Is a fraud finding the same as a recovery?**

No. New York insurers reported $3.5 billion in SIU-attributed savings for 2024 and $40.3 million in recoveries from SIU investigations in the same filings, so recoveries were 1.2% of claimed savings. Savings means money not paid out. Recovery means money clawed back, and it needs a different evidence package.

The pair of numbers that settles this sits in the same New York report. Insurers reported "approximately $3.5 billion in savings and approximately $40.3 million in recoveries from SIU investigations in 2024 (the most recent year for which data is available)". Derived, recoveries were 1.2% of claimed savings.

**Figure 3 (Savings against recoveries, 2024) - A fraud finding and a recovery are two different evidence packages.** Same report, same year, same insurers: what the industry counts as saved and what it actually got back are different numbers.

**1.2%** - **Recoveries as a share of claimed savings** New York insurers reported approximately $3.5 billion in savings and approximately $40.3 million in recoveries from SIU investigations in 2024, in the same electronically filed Annual SIU Reports.

|  | What it proves | Evidence it must carry | Decision it supports |
| --- | --- | --- | --- |
| Referral package (10 CCR 2698.38 fields) | Reasonable belief of fraud | Claim data, public records, synopsis | What gets referred to the state |
| Recovery package | Entitlement to be paid back | Liability, policy language, ledger | Pursue or waive, and the demand |

**Takeaway:** The referral package proves suspicion. A recovery has to prove entitlement, and that is a different file.

Source: New York DFS Consumer Protection and Financial Enforcement Division annual report, updated April 15, 2026, quoting insurers' electronically filed Annual SIU Reports for 2024, the most recent year for which data is available. The 1.2% is derived from the two reported figures. Referral contents per 10 CCR 2698.38 via the Coalition Against Insurance Fraud's California SIU regulation page.

Civil recovery is where the money is, and it needs a different package. Subrogation, salvage, rescission or recoupment all require liability facts, the policy language that creates the right, the payment ledger, proof of damages, and a demand a defending adjuster or arbitrator will accept.

Nothing in a lead package produces that second package, and the industry rarely builds it: [one in four US property-liability insurers records no salvage or subrogation recovery at all](/blog/subrogation-recovery-ai/) in a given year. That is why [every file gets screened for subrogation and salvage](/platform/subrogation-recovery/) rather than the ones somebody remembered to look at.

## One evidence pass, or one product per decision

**Why does platform architecture matter on a fraud-suspected claim?**

Because a fraud-suspected claim needs the same facts six times over: for triage, for the referral, for the coverage position, for the reserve, for the settlement and for the recovery demand. One evidence pass that serves all six is a different architecture from one product per decision, each rebuilding the file it needs.

On a suspicious claim, the facts that justify the referral are the same facts that decide the coverage position, set the reserve, price the settlement and support the recovery demand. One investigation-grade evidence pass that serves all of those decisions is built differently from a set of per-decision products, each re-reading the file.

That is the claim Hesper is built on, and the supporting figures are internal benchmarks. Manual investigation runs 14+ days per case and one investigator carries 200+ cases, which is why about 25% of flagged claims get a real workup. Running 15+ investigation phases in parallel takes a case to minutes, not weeks, and moves coverage of flagged claims from about 25% to 100%.

A unit that can work up every flag makes the accept-or-decline call on evidence rather than on capacity. Fraud detection is built in, so Hesper runs standalone, and it is complementary where a carrier already runs FRISS, Shift or Verisk. The head-to-head view is at [Hesper against Shift Technology](/compare/hesper-vs-shift-technology/).

Laid out stage by stage, the question becomes how many handoffs a carrier is willing to own. In the table, "outside the stated scope" is a statement about one named product's published scope, not a capability judgement about a company. The Shift column is the Lead Package Agent specifically, not Shift Technology.

| Stage of a fraud-suspected claim | What the stage requires | Detection layer: FRISS, Verisk, Shift Fraud & Risk | Shift Lead Package Agent | Hesper AI |
| --- | --- | --- | --- | --- |
| 1. Alert raised | A score or rule fires on a claim | This is the layer: scores and flags | Starts here, from an alert something else raised | Built-in detection, or ingests an existing vendor's flag |
| 2. Referral triage and acceptance | The SIU decides whether the flag is worth a case. Industry acceptance ratio for automated referrals: 36% | Produces the flag and can assign the case; whether to accept it is a human SIU call | Raises the quality of what the SIU sees, so more flags survive triage | Investigates every flag instead of triaging most of them away: about 25% to 100% coverage, a Hesper internal benchmark |
| 3. Package assembly | Claims data, public records, credentialing, adverse media, chronology, factual synopsis, party identifiers (10 CCR 2698.38) | Not this layer | This is the layer: "claims data, public records, credentialing checks, and adverse media" | 15+ investigation phases in parallel, producing an audit-ready file |
| 4. Finding | Somebody concludes what happened and can defend it | A score is not a finding | Shift's words: "a senior reviewer still decides what gets referred and how it moves forward" | A reasoned finding with sources; the investigator keeps decision authority and can override |
| 5. Mandatory report | SFC, FD-1 or eFD-1 within 60 days of reasonable belief (Cal. Ins. Code 1872.4); New York plan report by March 15 | Outside the stated scope | The package is the raw material for the report | Audit-trail native output that feeds the filing |
| 6. Coverage position and reserve | Policy and endorsement reading, reservation of rights, reserve movement | Outside the stated scope | Outside this agent's stated scope | Cited coverage positions, drafted reservation-of-rights letters, reserve recommendations |
| 7. Disposition: pay, partial, deny, negotiate, settle | Runs on its own statutory clock regardless of the referral (Fla. Stat. 627.70131(7)(a): 60 days to pay or deny) | Outside the stated scope | Outside this agent's stated scope | Demand-package digestion, benchmarked valuation, drafted responses, early litigation-risk flags |
| 8. Recovery | Subrogation, salvage, restitution, civil recoupment. New York 2024: $40.3 million recovered against $3.5 billion in claimed savings | Outside the stated scope | Outside this agent's stated scope | Every file screened for subrogation and salvage; demands drafted |

Four questions turn a demo about assembly speed into an evaluation about the whole claim.

1. Where does this agent's scope end, in your own words, and which of our downstream decisions does it not touch?
2. Does the coverage position draw on the same sourced record as the fraud finding, or does somebody read the policy again from the beginning?
3. What happens to the pay-or-deny clock while the referral is being prepared?
4. Who screens this file for subrogation and salvage, at what point, and against what evidence?

Automating the referral package well is a real contribution. Buy the assembly. Then ask what happens on day two.

## Key takeaways

- A referral package is a regulated artefact: 10 CCR 2698.37 and 2698.38 prescribe its contents, and California Insurance Code 1872.4 sets a 60-day filing deadline.
- New York's Insurance Frauds Bureau received 51,419 reports of suspected fraud in 2025 and opened 243 cases for investigation, roughly one report in 212.
- Automated tools generate 45% of all SIU referrals but only 29% of accepted ones: their acceptance ratio is 36% against 70% for adjuster referrals.
- Filing a referral does not stop the claim clock: Florida still requires the insurer to pay or deny a property claim within 60 days of notice under Fla. Stat. 627.70131(7)(a).
- New York insurers reported $3.5 billion in SIU-attributed savings for 2024 against $40.3 million in recoveries. Proving suspicion and proving entitlement are two different evidence packages.

## Frequently asked questions

### Does referring a claim to SIU stop the claim clock?

No. The two clocks run in parallel. Florida Statute 627.70131(7)(a) requires an insurer to pay or deny a property claim within 60 days of notice. California separately gives the carrier 60 days from reasonable belief to file the referral. A faster referral buys the investigative team working days. It does not buy the adjuster an extra day on the coverage decision, the reserve or the disposition.

### What happens after a fraud referral is made?

Very little at the state level. New York's Insurance Frauds Bureau received 51,419 reports of suspected fraud in 2025 and opened 243 cases. Meanwhile the carrier still has to take a coverage position, set the reserve, decide whether to pay, deny or negotiate, defend that decision, and pursue any recovery. The referral does not resolve the claim.

### Can AI write an insurance fraud referral package, and is that enough?

Yes to the first. Shift Technology's Lead Package Agent pulls together claims data, public records, credentialing checks and adverse media. But it ends, in Shift's own words, where a senior reviewer still decides what gets referred. The coverage position, reserve, disposition and recovery are separate work a referral package does not produce.
