110 incidents on record · 2026 Headlights Incident reports by Ellie Harris · Melbourne
10 new this week Library last updated 27 July 2026
← The incident library
HD-INC-103
Education · United States · 2024 · Overstated capability and procurement failure

Los Angeles schools committed about six million dollars to an AI chatbot showcased in spring, and its vendor had collapsed by July

By Ellie Harris · Filed Contract 2023 to 2024; 'Ed' showcased spring 2024

Alleged: Los Angeles Unified School District; AllHere Education developed or deployed the AI system implicated in this incident. Details are drawn from public reports; parties are presumed innocent of any wrongdoing not established by an official finding.

Los Angeles schools committed about six million dollars to an AI chatbot showcased in spring, and its vendor had collapsed by July

What happened

It was reported that the Los Angeles Unified School District, the second-largest school district in the United States, signed a contract worth up to about six million dollars with a company called AllHere to build “Ed”, an AI chatbot meant to help students and families with academic and wellbeing support, and was later reported to have paid roughly three million of that before the project unravelled. In the spring of 2024 the district’s superintendent, Alberto Carvalho, presented Ed at the ASU+GSV Summit, a major education-technology conference, describing it in ambitious terms. Within months the picture changed, because in June 2024 AllHere furloughed most of its staff, citing financial trouble, and the company filed for Chapter 7 bankruptcy.

According to reporting by The 74, a whistleblower who had worked at AllHere alleged that the company had mishandled students’ personal information while building the tool. When the vendor collapsed, questions followed about where the student data it held had gone. The company’s founder, Joanna Smith-Griffin, left the company, and in November 2024 federal prosecutors in New York charged her with securities fraud, wire fraud and aggravated identity theft, alleging she had misled investors about the company’s revenue and customers, including false claims that it held contracts with other school districts; she is entitled to the presumption of innocence unless convicted. The district said student data had been protected, but the episode left a publicly funded AI project defunct within months of its launch.

What an auditable version would have shown

When a public body hands a vendor both a large budget and children’s personal data, two things should be answerable from the start: whether the tool could actually do what was claimed, and where the data went. An auditable version records the evidence behind the capability claims before the money is committed, and keeps a standing record of what student data was collected, where it was stored, who could reach it, and what happens to it if the vendor fails. With those records, the collapse of a supplier becomes a controlled handover rather than a scramble to work out what data existed and who now holds it.

Where the gap was

A strong claim about an AI product, and a large amount of sensitive student data, appear to have been largely taken on trust, and when the vendor ran into trouble there was little clear record standing behind either. A VerificationGate holds a capability claim to evidence before a contract is signed, so “transformative AI platform” has to be shown rather than announced. An EgressGate governs student data leaving the district for a third-party service, recording what was sent and under what terms and requiring that it can be retrieved or destroyed on demand, so a supplier’s bankruptcy does not become a data black hole. The failure was not that a school district tried AI, but that neither the claim nor the data had a record standing behind it.

What governance should have looked like

Public money and children’s data raise the bar. Before a district buys an AI tool, it should require evidence that the tool works rather than a demonstration, and it should know exactly what data the vendor will hold and be able to get it back or have it destroyed if the vendor disappears. The lesson of Ed is that an AI procurement is only as safe as the records behind it, and that “the vendor collapsed” should never be the first time anyone asks where the data went.

Failure Pattern: a public body committed significant funds and sensitive student data to an AI product on the strength of unproven claims, with no durable record of the tool’s capability or of where the data went when the vendor collapsed.

Governance Principle: before a public body deploys an AI tool, the capability claim should be backed by evidence and the handling of any personal data should be recorded and recoverable, so a vendor’s failure does not put money or data beyond account.

The reference implementation of VerificationGate and EgressGate is open source. It lives at github.com/saffronandindia/headlights-oss, Apache 2.0 licensed and free to install. The repository is public now.

Sources

The mailing list

Fresh incident reports every week. One email to match.

We add new incidents to the library regularly, and send a single short email each week with what's new. The library stays free and open; this is just how you keep up with it.

No tracking. Unsubscribe in one click.

The record

An auditable system would have produced a signed, tamper-evident record the moment this happened: what the system did, the version that did it, the basis it acted on, and the action taken, and Los Angeles Unified School District; AllHere Education could have produced it on demand.

This is the record the system as deployed did not produce in a signed, auditable form.

What this teaches
Capture what happened when it happens
What the system did, the version that did it, the basis it acted on, and the action taken, recorded at the moment, not reconstructed after.
Sign it, so no one has to trust the record-keeper
A tamper-evident entry. Edit it later and the signature breaks. The record does not ask for the benefit of the doubt.
Make it verifiable by anyone
A court, a regulator, a customer's lawyer can check the record themselves, without taking the company, or us, at our word.

Headlights summarises publicly reported AI incidents. All summaries are independently written, attributed to their original sources, and intended for research and educational purposes. Allegations are identified as such until established through official findings.

Last reviewed June 2026. This report is based on the sources listed above and reflects information available at the time of review; later developments may not be captured. Where a person is described as charged with or alleged to have done something, that allegation is unproven unless a conviction or a court or regulatory finding is stated. Headlights publishes journalism and commentary, not legal advice.

Want to write back?

Direct to my inbox.

ellie@useheadlights.com →