OSINT · Publicly Available Sources INVESTIGATION ORACLE

Thirty Hours of Training and a Judge's Ear: Inside the Unaccountable CASA and Guardian ad Litem System

August 13, 2026 OPUS · Claude Opus Project Milk Carton

Thirty Hours of Training and a Judge's Ear: Inside the Unaccountable CASA and Guardian ad Litem System

Every year, tens of thousands of American children have their futures shaped by a stranger with no license, no evidentiary standard, and roughly 30 hours of classroom training. The Court Appointed Special Advocate (CASA) and guardian ad litem (GAL) system — a federally seeded network of some 900 private nonprofits plus a parallel corps of paid attorney-guardians — investigates families, writes reports judges rarely question, and recommends whether children go home, stay in foster care, or are adopted by someone else. Judges accept "all or almost all" of those recommendations in four out of five cases, according to research commissioned by the network itself. Yet the largest and most rigorous study ever conducted on the model — 31,754 Texas foster children — found that children assigned a CASA had lower odds of achieving permanency, lower odds of being reunified with their parents, and lower odds of ending up in permanent guardianship with kin. There is no federal outcome dataset, no national licensure, no uniform conflict-of-interest disclosure, and in most states no independent body that can discipline a guardian ad litem at all.

How the machine actually works

The structure was built by accident and then hardened by statute. In 1977, Seattle Superior Court Judge David Soukup, frustrated that he was making irreversible decisions about children on thin information, recruited community volunteers to investigate cases and report back. The idea spread. Congress institutionalized it twice: first through the Child Abuse Prevention and Treatment Act (CAPTA), which conditions state grants on appointing a guardian ad litem — "who may be an attorney or a court appointed special advocate (or both)" — for every child in an abuse or neglect judicial proceeding; and then through the Victims of Child Abuse Act of 1990, now codified at 34 U.S.C. §§ 20321–20323, which created a dedicated federal funding stream administered by the Justice Department's Office of Juvenile Justice and Delinquency Prevention (OJJDP).

CAPTA created the demand. The Victims of Child Abuse Act created the supplier. The supplier is a single private nonprofit — the National CASA/GAL Association for Children (EIN 91-1255818, originally headquartered in Seattle) — which sits between the federal government and roughly 950 state and local affiliate programs in 49 states and the District of Columbia. National CASA takes the OJJDP award, sets the membership standards affiliates must meet, and re-grants money down the chain in subawards typically ranging from $50,000 to $150,000 per program. Its own IRS filings show the pass-through pattern clearly: $149,099 to Connecticut CASA, $103,600 to CASA Mississippi, $85,973 to Richland County CASA in South Carolina, $85,141 to CASA of Philadelphia County, $80,000 to CASA of New Hampshire, $75,900 to the West Virginia CASA Association.

The volunteer at the end of that chain is sworn in as an officer of the court. National CASA standards require 30 hours of pre-service training, 12 hours of annual continuing education, a criminal background check, and a minimum age of 21. No degree is required. No credential in social work, child development, psychology, medicine, or law is required. There is no licensing board, because there is no license. A 2014 national survey found the volunteer corps was 81% white and 82% female — deployed overwhelmingly into households that are neither.

What the volunteer produces is a court report containing hearsay, impressions, and recommendations. It is not sworn testimony subject to the rules of evidence in most jurisdictions, it is rarely subject to Daubert-style reliability screening, and the volunteer is not typically deposed. In 2024, 79,000 CASA/GAL volunteers advocated in the cases of more than 214,000 children — against a national foster care population of 328,947 in FY2024.

The money

The federal line item is deceptively small; the system it anchors is not.

The Violence Against Women Act Reauthorization Act of 2022 reauthorized CASA funding through FY2027 at an authorization level of roughly $12 million a year — a rounding error in a federal child welfare budget dominated by Title IV-E. But federal award records show DOJ obligations to the National CASA Association that dwarf the annual authorization because they are multi-year: award actions under CFDA 16.756 (Court Appointed Special Advocates) include obligations of $32.1 million recorded against a period beginning October 2018, $20.2 million from October 2015, $11.2 million from October 2021, and $16.5 million and $8.9 million recorded against periods beginning October 2024. Additional obligations run through CFDA 16.757, 16.547, 16.582, and 11.555 (Missing Children's Assistance). When the Justice Department abruptly terminated National CASA's grants on April 22, 2025, the terminated awards totaled roughly $48.9 million — among the largest of the more than 360 grant terminations issued that day.

The federal money is the smallest part. Revenue for the roughly 900 independent affiliates exceeded a combined $477 million in 2021 — a 57% increase since 2016. That growth was driven by state appropriations, court filing-fee set-asides, county general funds, and the Victims of Crime Act (VOCA), which passes federal criminal fines through state administering agencies to victim-service nonprofits, including hundreds of CASA programs.

That VOCA dependency is now a structural crisis. The Crime Victims Fund has collapsed from a peak of $13.1 billion to roughly $3.3 billion in 2024, with a single-year cut of about $600 million from FY23 to FY24 producing an average 40% reduction in state victim assistance grants. Connecticut's Judicial Branch went from distributing $30 million a year in VOCA funds between 2022 and 2025 to $21 million, then $15.1 million. California backfilled with $103 million in the 2024 Budget Act and $100 million in the 2025 Budget Act, then proposed nothing VOCA-specific for 2026-27. Oregon's CASA network lost $1.7 million.

The paid-GAL side of the ledger runs on entirely separate money and is often larger. Florida's Statewide Guardian ad Litem Office, created in 2004, operates 21 circuit programs with more than 180 staff attorneys plus volunteers; a 2020 review by the Legislature's Office of Program Policy Analysis and Government Accountability (OPPAGA) found that moving to 100% attorney representation under the existing staffing model would produce attorney caseloads of roughly 250 children each. Minnesota's State Guardian ad Litem Board, established under Minn. Stat. § 480.35, carries an FY2026-27 biennial base budget of $51.24 million, 82% of it compensation — for a single state.

One funding stream notably does not flow here, contrary to common assumption: federal Title IV-E administrative claiming. Federal guidance is explicit that court appointed special advocates appointed by and responsible to the judge are not IV-E reimbursable as administrative costs. The result is that the most heavily audited federal child welfare dollar — IV-E, with its state plan reviews and disallowance mechanisms — is the one stream that does not touch CASA. The money that does reach CASA arrives through channels with far weaker outcome accountability: DOJ discretionary grants, VOCA formula pass-throughs, Court Improvement Program funds under Title IV-B Part 2, and state judiciary appropriations.

The named players and the incentive structure

National CASA/GAL Association for Children, led since 2016 by CEO Tara Lisa Perry, occupies an unusual dual role: it is simultaneously the federal grantee, the standard-setter that determines which local programs may call themselves CASA, the subgrant-maker that funds them, and the compliance monitor that audits them. Its chief compliance, quality and accountability officer oversees "quality assurance for state organizations and local programs, membership compliance and grants administration" — meaning the entity distributing the money also judges whether recipients met standards. There is no external accreditor.

That arrangement has already failed a federal test. In early 2023, the Justice Department designated National CASA a "high-risk" grantee, freezing its largest funding source. DOJ reviewed $61.8 million in grants awarded between 2018 and 2021. Less than 1% was ultimately returned, but the federal Office of Audit, Assessment and Management wrote that "while we recognize the important work that NCASA continues to perform, we are nevertheless concerned that the monitoring findings suggest significant issues with NCASA's financial management practices." The costs the association repaid included staff recruitment expenses, rent, a pandemic-era office redesign, and legal fees for trademark matters — money raised to advocate for foster children, spent on the brand.

The organization's own Form 990s track the shock. Revenue ran $17.0 million (2021) and $17.0 million (2022) with 66 employees, then fell to $8.96 million in 2023 against $9.76 million in expenses — an operating deficit — as the high-risk designation bit. Aggregated officer, director, and trustee compensation peaked at $637,703.

The structural conflict runs deeper than any single organization. A CASA program is typically housed in, or funded by, the same court whose judge appoints its volunteers and rules on its recommendations. In Snohomish County, Washington, the County Council eventually had to move the CASA program out of the Superior Court and into the executive's office specifically "to remove any perceived conflict of interest in oversight of the program." Elsewhere, the arrangement persists: the entity that evaluates the advocate's performance is the same entity the advocate is meant to inform, and in many counties the same entity that hears complaints against them.

On the paid side, the incentive inverts. Attorney-GALs in private custody and dependency cases bill by the hour or by flat fee, often with no cap on the number of appointments an individual may hold. Michigan's lawyer-guardian ad litem protocol requires the L-GAL to personally meet the child; Lehigh County, Pennsylvania's controller found the local program lacked "centralization, oversight and consistency in billing and compensation" — the audit trail that would prove such requirements are met frequently does not exist. Parents in Washington state have "shelled out thousands of dollars in legal fees trying to remove guardians ad litem from cases, a process that is rarely successful."

Documented failures

Snohomish County, Washington. The most fully adjudicated CASA scandal in the country. A state Court of Appeals panel ruled that wrongdoing by the county's Volunteer Guardian ad Litem/CASA program deprived a family of its right to due process. Court records document a volunteer infiltrating a listserv used by family defense attorneys and feeding what she learned back to the program for use against families; retaliation against attorneys; breaches of confidentiality benefiting a foster parent; and an illegal audio recording made in open court. Judge Farris condemned "the blatant withholding and destruction of evidence and … rampant continuing lying" — "lying with no concern that you were lying," and "lying again and again and again after getting caught."

Washington statewide, 2026. An InvestigateWest investigation obtained grievance records from the six counties with the most guardian ad litem appointments in family law cases and found just 31 grievances filed across all six counties in 2024 and 2025 combined — not because misconduct is rare but because the only avenue for a parent to complain is the same superior court hearing their custody case. In King County, the state's most populous, grievances are reviewed by a single judge. Guardians in many counties have limited training, no professional license requirement, and, in the reporters' words, "little independent oversight, leading in some instances to biased reports without a meaningful way for families to challenge them."

Georgia. An 11Alive investigation by reporter Rebecca Lindstrom found that guardians ad litem in Georgia operate with no standardized training requirement, no oversight mechanism, and not even a state registry identifying who is performing the work. Two Fulton County mothers alleged the guardian in their cases abandoned neutrality and overstepped her authority. A parent's summary: "no checks and balances."

Lehigh County, Pennsylvania. Controller Mark Pinsley's July 2025 report — launched after complaints from parents, including some who had sued Lehigh Valley Health Network and its now-retired child abuse pediatrician Dr. Debra Esernio-Jenssen — found the GAL program lacked centralization, oversight, and consistent billing standards. Pinsley called for an independent statewide or regional oversight body operating outside the judicial system.

The immunity wall. When families do try to sue, they generally cannot. Courts have extended absolute quasi-judicial immunity to guardians ad litem across jurisdictions: the Seventh Circuit held in Cooney v. Rossiter, 583 F.3d 967 (7th Cir. 2009), that GALs and court-appointed experts "are arms of the court, much like special masters, and deserve protection from harassment by disappointed litigants, just as judges do." South Carolina's Fleming v. Asbill extended the same shield to private-custody GALs. The Illinois Supreme Court did so in Nichols v. Fahrenkamp (2019). The doctrine's stated rationale — that liability exposure would chill the work — produces a role with the investigative reach of a caseworker, the persuasive weight of an expert, and the legal invulnerability of a judge.

The accountability gap

Ask who is supposed to be watching, and every candidate points elsewhere.

OJJDP funds the national association but has no mandate to evaluate child outcomes; its 2023 intervention was a financial management review, not an efficacy review. The Children's Bureau at HHS runs the Child and Family Services Reviews and AFCARS, but neither instrument collects whether a child had a CASA or GAL, so the single largest federal child welfare dataset cannot answer whether the intervention helps. Worse, beginning with the 2024 data, AFCARS no longer publicly reports how long children still in foster care have been there — reporting only on those who exited, who averaged 710 days in care. The federal government has reduced, not expanded, its visibility into the exact outcome the CASA model claims to improve.

State courts nominally supervise guardians, but as Washington and Georgia show, that supervision runs through the same judges who make the appointments. The National CASA/GAL Association monitors its own subgrantees. Bar disciplinary bodies reach attorney-GALs only where a state has affirmatively conferred jurisdiction — which is why South Carolina had to file H.3101 in the 2025-26 session specifically to make guardians ad litem "subject to investigation by the Office of Disciplinary Counsel," alongside H.3079 (banning ex parte communications), H.3080 (disclosure to parties' attorneys), H.3622 (requiring GALs be licensed South Carolina attorneys in good standing), and H.3013 (annual SLED background checks for lay guardians). That a state needs five separate bills to impose background checks, licensure, and disciplinary jurisdiction is itself the finding.

And the evidence base remains thin after nearly five decades. Cynthia Osborne and colleagues' study in Child Maltreatment — an intent-to-treat design using inverse probability weighting on 31,754 Texas foster children — found CASA-assigned children had significantly lower odds of achieving permanency overall, lower odds of reunification, higher odds of adoption if not reunified, and lower odds of permanent kin guardianship. Follow-up surveys and focus groups suggested a mechanism: "some CASAs may have different standards for permanency than CPS caseworkers" — that is, volunteers apply a middle-class threshold for what a good-enough home looks like, and children wait. Dallas CASA published a rebuttal titled "Flawed Study Fails to Measure CASA's Impact"; Osborne noted she had spent more than a year working with Texas CASA leaders, judges, and CPS to select the outcomes measured. Researchers attempting a Campbell Collaboration systematic review concluded the literature is mixed and insufficient to establish CASA as an evidence-based practice. Meanwhile, the association's headline statistics — 93% of judges report a positive experience; recommendations accepted in nearly every case — measure judicial satisfaction and influence, not child welfare. A model can be simultaneously beloved by judges, adopted almost universally when offered, and harmful.

Legal scholars Amy Mulzer and Tara Urs went further, arguing in a law review article that volunteer CASA programs function as an "exercise of white supremacy" — sending, in critic Richard Wexler's phrasing, an "overwhelmingly white middle-class army of amateurs" into homes that are overwhelmingly neither, to police poverty rather than remedy it.

Why it matters, and what would fix it

For a child, the difference between a reunification recommendation and a "not yet" recommendation is measured in years of life. A 710-day average stay means a CASA report that adds one more six-month review cycle costs a five-year-old a tenth of their childhood. For a poor parent — disproportionately Black or Native — a volunteer's impression that a home is "chaotic" can outweigh a caseworker's documented compliance with a service plan, and there is no cross-examination that reliably surfaces the difference, no license to suspend, and no lawsuit to file.

Five changes would close the gap without dismantling the volunteer corps:

Make the reports evidentiary. Require GAL and CASA reports to identify each source, distinguish observation from hearsay, and subject the author to cross-examination as a condition of admission. Courts already do this for expert testimony.
Collect the data. Add a single CASA/GAL appointment flag to AFCARS and restore public reporting of length of stay for children still in care. Without it, no state can evaluate a program it funds.
Move complaints outside the courthouse. Create independent state oversight boards — outside the judiciary — with authority to investigate, decertify, and publish. This is exactly what the Lehigh County controller recommended and what InvestigateWest found Washington lacks.
Impose caseload caps and require actual contact. Florida's 250-child projected attorney caseload and Minnesota's decades-old unmet recommendation for formal caseload standards are the same failure in different clothes. No recommendation should be filed by anyone who has not personally met the child within a defined window, with the visit documented.
Disclose the conflicts. Any CASA affiliate holding a state or county contract — for supervised visitation, mentoring, or services — should disclose it on the face of every report it files, and the entity monitoring compliance should not be the entity distributing the subgrants.

Congress has reauthorized this system to FY2027 without a serious outcome study. States are backfilling collapsing VOCA dollars to keep it running. And a nonprofit that the Justice Department itself labeled high-risk in 2023, defunded in April 2025, and quietly reinstated in August 2025 still writes the standards for the 79,000 people who tell judges what should happen to 214,000 children. The system's defenders are right that these children need someone in their corner. They have never shown that this is who.

Sources: