The Signature That Replaces a Judge: Inside America's Hidden Foster Care System

Every year, somewhere between 100,000 and 300,000 American children are moved out of their parents' homes by child protection agencies without a judge ever signing an order, without a lawyer ever being appointed, and without the child ever entering the legal custody of the state. The instrument is not a warrant — it is a "safety plan" or a "kinship placement agreement," a one-page document a caseworker presents to a frightened parent at the kitchen table with an unspoken alternative: sign, or I file a petition and your child goes to strangers. Because the agency never takes custody, the child never appears in the federal Adoption and Foster Care Analysis and Reporting System, never gets a permanency hearing, never triggers a dollar of Title IV-E maintenance payments, and never generates a case a court can review. Legal scholars call it "hidden foster care." The federal government does not call it anything, because the federal government does not count it. In FY 2024, the official U.S. foster care population was 328,947 children — a number that may be understating the true scope of state-facilitated family separation by half or more.

The Setup: How a Signature Replaces a Judge

The mechanism is elegant in its avoidance of law. When a caseworker investigating an abuse or neglect report concludes a child should not stay in the home, the statutory path is to petition a juvenile or family court. That petition triggers a cascade of protections built into federal law over four decades: a judicial determination that remaining home is "contrary to the welfare" of the child, a finding that the agency made "reasonable efforts" to prevent removal under 42 U.S.C. § 671(a)(15), appointment of counsel for the parent in most states, a guardian ad litem for the child, a written case plan under 42 U.S.C. § 675(1), and a permanency hearing within twelve months under § 675(5)(C).

Hidden foster care skips all of it. The caseworker instead secures the parent's "voluntary" agreement that the child will live with a grandmother, an aunt, a family friend. No petition is filed. No judge is involved. And critically — no home study, no licensing inspection, and often no completed background check on the household receiving the child.

The federal statute itself creates the incentive. Under 42 U.S.C. § 672(a)(2)(A), federal foster care maintenance payments flow only when a child is removed pursuant to a judicial determination or a formal voluntary placement agreement, and only when the title IV-E agency holds "placement and care responsibility." The AFCARS reporting regulations at 45 C.F.R. § 1355.41 define the reportable population the same way. So the very act of keeping a case out of court accomplishes two things simultaneously: it strips the family of due process, and it erases the child from the federal count. The child is separated in fact but not in law. Nobody owes them anything.

The practice operates under a dozen euphemisms — "kinship diversion," "foster care diversion," "safety planning," "informal family arrangement," "parental placement." The Hidden Foster Care Transparency Act, now pending in the 119th Congress, defines it in plain language: "any separation of a child from the child's parents or primary caregivers that occurs without the State taking responsibility for the care or placement of the child and without a court order or the involvement and oversight of a court of law, whether voluntary or involuntary."

That last clause — whether voluntary or involuntary — is the entire fight.

The Numbers That Don't Exist

Professor Josh Gupta-Kagan, now at Columbia Law School, published the foundational analysis in 2020: America's Hidden Foster Care System, 72 Stan. L. Rev. 841. His conclusion was that agencies induce these transfers "tens and likely hundreds of thousands of times annually," and that the practice is "hidden from courts because agencies file no petition alleging abuse or neglect and from policymakers because agencies do not generally report these cases."

Six years later, that remains true. A 2022 survey conducted by Child Trends for the Annie E. Casey Foundation — sent to all 50 states, D.C., and Puerto Rico, with 46 jurisdictions responding — found that 33 states have policies permitting kinship diversion. Roughly half of those states claim to track diversion arrangements internally, but almost none can report publicly how many children were diverted or what happened to them. Only two states require any court oversight of a diversion placement.

The scale estimates that do exist come from extrapolation, not counting. Child Trends and Casey put the range at 100,000 to 300,000 children per year. In some jurisdictions studied, for every ten children formally entering foster care, another seven were diverted; in others the split was one-to-one. Duration data is equally sparse, but where it exists it demolishes the "temporary emergency" framing: about 60% of diversion arrangements last under 90 days, while nearly one in four last more than six months. Six months is longer than many formal foster care episodes — and it passes without a single hearing.

Set that against the official ledger. AFCARS reported 328,947 children in care on September 30, 2024, the sixth consecutive annual decline from a 2020 peak of 407,332. About 39% of those children — roughly 127,000 — were placed with relatives or kin inside the system, with all the licensing, payments, and oversight that entails. The children in hidden foster care are placed with the same category of relatives, in the same circumstances, for the same reasons — and receive none of it.

Here is the uncomfortable implication: the celebrated six-year decline in America's foster care population may be, in part, a measurement artifact. If agencies increasingly route children through a channel that federal data does not capture, the caseload numbers fall without a single additional child staying home.

The Money: Diversion Is a Cost-Avoidance Strategy

Follow the dollars and the practice stops looking like a child-welfare philosophy and starts looking like a budget maneuver.

When a child enters formal foster care, the state must pay a foster care maintenance payment and can claim roughly half of it back from the federal government under Title IV-E. When a child is diverted, the state pays nothing. The caregiver — typically a grandmother on a fixed income — is directed to apply for a TANF child-only grant, a program funded from the state's TANF block grant with no federal foster care match.

The gap is not marginal. It is a chasm:

• The average monthly foster care maintenance payment supporting a relative foster parent is approximately $915 per child.
• TANF child-only grants start as low as $81 per month for the first child. The national average is about $328; state rates range from $60 to $514, with a median of $215.
• In Texas, the daily kinship reimbursement is $12.67 per child against $27.07 for a licensed foster family — about $4,625 per year versus $9,881.

A state that diverts a child instead of placing them saves the difference, every month, and shifts the remainder onto a caregiver who never agreed to become a foster parent and cannot afford to be one. Diversion is, functionally, an unfunded mandate imposed on grandmothers.

Meanwhile, the money the system does spend is not going to children. A February 2026 issue brief from HHS's Office of the Assistant Secretary for Planning and Evaluation, Administrative Costs Drive Foster Care Claims, found that from FY2019 to FY2023, administrative costs — not payments to foster parents — were the single largest driver of Title IV-E foster care expenditures. In-placement administration alone exceeded every other expense category by more than $1 billion. Per child in FY2023, states reported average monthly expenditures of $2,576 on in-placement administration, $869 on pre-placement administration, and $530 on other administration.

Read that against the $915 maintenance payment and the $215 median TANF grant. The federal government reimburses roughly $2,576 a month to administer a child's case and $915 to feed and clothe them — and a diverted child gets $215 and no case at all. ASPE further found that average monthly administrative expense per child rose 40% from FY2019 to FY2023 while the number of children served declined, and that of pre-placement administrative claims, 97 to 99% went to case planning and management while only one to three percent went to legal representation — the very function that would put a lawyer next to the parent at that kitchen table. The federal cost structure is now the subject of a proposed rule published in the Federal Register on July 30, 2026, Administrative Costs for Children in Title IV-E Foster Care.

Georgia is a useful illustration of the sums involved. Federal foster care funding tracked to the state totals approximately $1.43 billion across six programs and 204 awards. Georgia recorded 11,435 substantiated child maltreatment victims and 103 child fatalities in 2023. None of that money reaches a diverted caregiver.

"Blatantly Coercive" or "A Martini or a Manhattan": The Courts Split

The legal question is whether a parent facing "sign this or I take your child" has given consent at all. Federal appellate courts have answered in opposite directions, and the split has never been resolved.

In Croft v. Westmoreland County Children and Youth Services, 103 F.3d 1123 (3d Cir. 1997), a caseworker acting on an uncorroborated anonymous tip told Dr. Henry Croft that unless he left his home that night and had no contact with his four-year-old daughter, the child would be placed in foster care. He left. The Third Circuit refused to call it voluntary, holding the threat "blatantly coercive" and the agency's conduct "an arbitrary abuse of governmental power," because the investigation had produced no objective evidence of abuse.

Nine years later, the Seventh Circuit went the other way in Dupuy v. Samuels, 465 F.3d 757 (7th Cir. 2006), a challenge to Illinois DCFS safety plans. Writing for the court, Judge Richard Posner reasoned that safety plans are "optional," "impose no obligations on anyone," and that a parent who is not abusing a child can simply refuse. "We can't see how parents are made worse off by being given the option of accepting the offer of a safety plan," he wrote, analogizing the choice to being offered a martini or a Manhattan. Coercion, in that framework, requires illegal means — a physical threat. The threat to invoke a legal process does not count.

The Sixth Circuit followed the Dupuy logic in Smith v. Williams-Ash, No. 06-4638 (6th Cir. 2008), affirming summary judgment for a Hamilton County, Ohio caseworker on the ground that parents who signed a "voluntary" safety plan had consented to the removal and were therefore owed no hearing. Notably, the defendant argued the plan was voluntary but never entered the document itself into the record.

The doctrine has begun to move. In K.W. v. City of New York, No. 24-3042-cv (2d Cir. 2026), the Second Circuit reinstated Fourth Amendment unlawful-seizure and Fourteenth Amendment procedural due process claims arising from a child removed without a court order. And in B.B. v. Hochul, a class action filed in 2021 in the Eastern District of New York by the Legal Aid Society and Dechert LLP on behalf of 14 New York City children denied placement with relatives, the Second Circuit reversed a standing-based dismissal in January 2026 and remanded for the 11 remaining plaintiffs. The panel's reasoning cuts to the heart of the harm: the children "suffered a real-world harm," some raised by strangers and exposed "to risks of psychological and emotional harms," and others pushed into informal relative caregiver arrangements that deprived them of "medical and social services provided to children in foster care." That is a federal appellate court describing hidden foster care as a cognizable injury.

Voters are moving too. On November 4, 2025, Texans approved Proposition 15, embedding a parental right to "exercise care, custody, and control" of children in the state constitution, with just under 70% of the vote and majorities in 253 of 254 counties. The measure — introduced by State Sen. Bryan Hughes, passed 31–0 in the Senate and 112–22 in the House — was amended before passage to strip the strict-scrutiny language that would have required state interference to be narrowly tailored to a compelling interest, which sharply limits its practical bite against safety plans.

Georgia: What Happens When Nobody Vets the Placement

The safety argument for hidden foster care is that kin placements are better for children than stranger care. That is well supported by research — for vetted kin placements, inside the system, with support. Diversion is a different thing: kin placement stripped of vetting and support.

Senator Jon Ossoff's Senate subcommittee ran a 13-month bipartisan investigation into Georgia's Division of Family and Children Services, launched in February 2023 and reported out in April 2024. Its findings included that DFCS failed to adequately assess and respond to safety risks in 84% of cases reviewed between March and June 2023; that nearly 2,000 children in the agency's care were reported missing between 2018 and 2022, with at least 410 likely sex trafficked; and that the agency retaliated against employees who reported internal wrongdoing.

Buried in that report is the specific indictment of diversion. Investigators found that "voluntary kin placements are not appropriately vetted," and that Child Advocacy Centers reported cases in which children were placed in the homes of people with histories of sex crimes or prior DFCS involvement. That is the predictable consequence of a placement process with no home study, no licensing standard, and no judge. DFCS responded with an 11-page rebuttal accusing the senator of "political gamesmanship."

The Ossoff investigation is what produced the federal legislation. S. 2902, the Hidden Foster Care Transparency Act, was introduced by Sens. John Cornyn (R-TX) and Jon Ossoff (D-GA) and referred to the Senate Finance Committee; the House companion, H.R. 5507, was introduced by Rep. Nathaniel Moran (R-TX). A predecessor, S. 5214, died in the 118th Congress. The bills would require states to report to HHS the number of children in hidden foster care arrangements, how those arrangements end, the allegations that produced them, whether the caregiver had legal counsel, and what services the caregiver received — and require HHS to publish it. At the state level, New York S5242 (2025) would impose parallel reporting on local CPS agencies.

The Caregiver's Trap

Consider what a grandmother actually receives when she says yes. She gets a child, often overnight, often with no clothes and no medical records. She does not get a foster care payment. She frequently cannot enroll the child in school or consent to medical treatment, because she has no legal custody and no court order conferring authority — a problem Casey Family Programs identifies directly. She is not eligible for the child-welfare services attached to a formal case. She is not represented by counsel, because there is no case. And she has no idea how long the arrangement lasts, because no statute or order sets a term.

The parent's position is worse. Having "voluntarily" placed the child, the parent has no right to a hearing to get them back, no case plan defining what compliance looks like, no reunification clock, and no lawyer. If they attempt to retrieve the child, the agency can and does file the petition it withheld — and cite the parent's "instability" as evidence. The Minnesota Journal of Law & Inequality captured the dynamic in its 2022 analysis, Signing Away Your Right to Parent: How Safety Plans Evade Due Process Requirements in Child Welfare Proceedings. The Texas Public Policy Foundation, from the opposite end of the political spectrum, published the same critique in The Hill under the heading "Shadow removals." This is one of the rare child welfare issues where the ACLU-adjacent family defense bar and conservative parental rights groups agree entirely.

The Accountability Gap

Every institution that should be watching is structurally excluded.

Courts cannot review a case that was never filed. Child welfare agencies have no monitoring obligation once they declare a case closed, so no one checks whether the child is safe or whether reunification is possible. The Children's Bureau audits states through Child and Family Services Reviews and AFCARS compliance — both of which measure only children in agency placement and care responsibility, which by construction excludes every diverted child. Congress appropriates against caseload numbers that omit them. State legislatures in 33 states have authorized the practice without requiring anyone to count it. Citizen review panels and foster care ombudsmen have jurisdiction over the formal system only.

The result is a population of children the size of the official foster care system about which the United States government holds essentially no data: not how many, not for how long, not with whom, not with what outcome, not how many were ever returned home.

What Would Actually Fix It

The reforms are known, and several are already federal law waiting to be used.

Count them. Pass S. 2902 / H.R. 5507. Reporting is not a solution, but every solution is downstream of a number.

Use the 2023 kin licensing rule. ACF's final rule on Separate Licensing or Approval Standards for Relative or Kinship Foster Family Homes, published September 28, 2023 and effective November 27, 2023 (ACF Information Memorandum IM-23-07), lets states adopt kin-specific licensing standards limited to federal safety requirements — and requires that licensed kin homes receive the same maintenance payment as non-kin homes. This eliminates the historical excuse for diversion, which was that grandma's house could never pass a licensing inspection written for strangers. Every state that still diverts while declining to adopt kin-specific standards is choosing to.

Fund the navigators. The Family First Prevention Services Act authorized 50% federal match for evidence-based kinship navigator programs. As of January 2026, only 11 states and Puerto Rico — Colorado, Delaware, Iowa, Minnesota, Nebraska, Nevada, Ohio, South Carolina, Utah, Virginia, and Washington — have approved programs. Thirty-nine states have left that money on the table.

Put a lawyer in the room. Title IV-E administrative funds can reimburse legal representation for parents and children, yet ASPE found only one to three percent of pre-placement administrative claims go to it. Pre-petition legal advocacy — the model the Barton Center and the Alliance for Children's Rights have promoted — is the single intervention most likely to convert a coerced signature into an informed choice.

Set a hard clock. No safety plan separating a child from a parent should run past a short statutory limit — 30 days is the common proposal — without either dissolution or a court petition. Two of thirty-three states require court oversight. That number should be thirty-three.

Pay the caregiver. A child living with a grandmother because a state agency put them there is a child in state-facilitated care, whatever the paperwork says, and should be supported accordingly.

Virginia's JLARC found in its 2018 foster care review that only 6% of the state's children in care were placed with relatives against a roughly 30% national average — while the state simultaneously ran one of the country's more aggressive diversion programs. That is the pattern in miniature: the relatives were there the whole time. The state simply preferred an arrangement in which it owed them nothing and told no one.

Hidden foster care is not a loophole in the child welfare system. It is a parallel system, roughly the same size as the visible one, built specifically in the space where federal money and judicial review both stop. Children enter it by a parent's signature and leave it — if they leave it — by nobody's order at all.

---

Sources

• Josh Gupta-Kagan, America's Hidden Foster Care System, 72 Stan. L. Rev. 841 (2020) · Stanford Law Review · SSRN
• H.R. 5507, Hidden Foster Care Transparency Act, 119th Congress · S. 2902 · Rep. Moran press release
• Sen. Ossoff — Findings of 13-Month Probe into Safety of Foster Children · Imprint: Senate Probe Finds 'Systemic Failures' · Georgia Recorder
• Annie E. Casey Foundation — New Insights on State Kinship Diversion Policies · The Kinship Diversion Debate · Child Trends — State Tracking of Kinship Diversion · Chapin Hall — Diverting Children from Foster to Kinship Care
• ASPE — Administrative Costs Drive Foster Care Claims (Feb. 2026) · Federal Register — Administrative Costs for Children in Title IV-E Foster Care (July 30, 2026)
• AFCARS FY2024 Dashboard (ACF) · ACF AFCARS data page · The Imprint — Number of Youth in Foster Care Declines Again in 2024
• Federal Register — Separate Licensing or Approval Standards for Relative or Kinship Foster Family Homes (Sept. 28, 2023) · ACF IM-23-07 · ACF Kinship Navigator Program
• Croft v. Westmoreland Cnty. CYS, 103 F.3d 1123 (3d Cir. 1997) · Dupuy v. Samuels (7th Cir. 2006) · Smith v. Williams-Ash, No. 06-4638 (6th Cir. 2008) · Grandfamilies — Hidden Foster Care and Kinship Diversion Caselaw
• B.B. v. Hochul (Legal Aid Society) · Imprint — Court OKs Lawsuit by Kids Who Were Denied Placement With Kin · K.W. v. City of New York (2d Cir. 2026)
• Grandfamilies — Uneven Support for Kinship/Grandfamilies: State TANF Child-Only Grants · Every Texan — Kinship Care in Texas · ASPE — On Their Own Terms: Supporting Kinship Care Outside TANF and Foster Care
• The Imprint — Hidden Foster Care series · Casey Family Programs — Hidden Foster Care · ABA — Addressing Hidden Foster Care · Alliance for Children's Rights
• Minn. J.L. & Ineq. — Signing Away Your Right to Parent · The Hill — Shadow removals · Barton Center — Guidance for Preventative Legal Advocacy
• JLARC — Improving Virginia's Foster Care System (2018) · NY S5242 (2025) · Texas Proposition 15 (2025), Ballotpedia
• Georgia federal foster care funding ($1.43B / 6 programs / 204 awards) and Georgia maltreatment statistics (11,435 victims, 103 fatalities, 2023): PMC CivicOps database — taggs_fostercare_congressional_districts, state_welfare_statistics

A note on verification: WebFetch was permission-blocked for several primary-source domains (congress.gov HTML, imprintnews.org, aspe.hhs.gov, grandfamilies.org, casey.org), so figures from those sources are drawn from search-result extracts rather than full-document reads. The core statistics — AFCARS FY2024 counts, the ASPE administrative-cost figures, the 33-state/2-state diversion policy counts, the 100k–300k estimate, the TANF-vs-FCMP payment gap, and the Ossoff findings — were each corroborated across multiple independent results. Two 990-lookup attempts (Casey Family Programs, Annie E. Casey Foundation) returned fuzzy-matched wrong entities, so no foundation financials are cited; the money analysis rests on Title IV-E and TANF, which is where the actual incentive lives. The FY2024 entry/exit counts and the Smith v. Williams-Ash reporter citation could not be independently confirmed — the former is attributed to the AFCARS dashboard, and the latter is cited by docket number only.