AllHere's Chatbot 'Ed' Fails and Costs Los Angeles Unified School District $6 Million

The Los Angeles Unified School District invested up to $6 million in developing the AI chatbot "Ed," meant to provide academic and mental health support for students. The chatbot allegedly failed to meet expectations, and the project collapsed when AllHere, the contracted start-up, faced financial difficulties. Prince George's County Public Schools in Maryland was also affected. Founder Joanna Smith-Griffin was removed as CEO in June 2024 and later arrested on fraud charges.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.3 Fraud, scams, and targeted manipulation

The founder engaged in a deliberate and targeted financial fraud scheme, misrepresenting the AI startup's success to deceive investors and school districts for personal financial gain.

Additional risk subdomains

• 2.1 Compromise of privacy by obtaining, leaking or correctly inferring sensitive information: A former software director alleged that the chatbot processed student records in violation of privacy rules, unnecessarily sharing PII with third-party companies.
• 7.3 Lack of capability or robustness: The sudden collapse of the company led to the abrupt shutdown of the Ed chatbot, leaving school districts without the contracted AI services and technical support.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The incident was caused by the intentional fraudulent actions of the company's founder, which occurred while the AI chatbot was actively deployed in school districts.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The AI system Ed is an educational tool used in public school districts to create individual learning plans and manage student data, which falls under educational systems affecting access to education.

AI system and alleged parties

• AI system: Ed (AllHere)
• AI purpose: Chatbot; Question Answering
• Behaviour type: Assistant
• Alleged developer: AllHere
• Alleged deployer: Los Angeles Unified School District (LAUSD), AllHere, Alberto Carvalho
• Alleged harmed parties: Taxpayers, Prince George's County students, Prince George’s County Public Schools, Prince George's County parents, Los Angeles Unified School District (LAUSD), Los Angeles students, Los Angeles parents

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Substantial. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Substantial, indirect Substantial
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Substantial, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: The reports explicitly describe significant financial losses, including nearly $10 million fraudulently obtained from investors and $3 million paid by LAUSD for a defunct chatbot.

Directly caused: Investors lost nearly $10,000,000 due to the fraud scheme, and LAUSD lost $3,000,000 paid to AllHere for a chatbot that was subsequently shut down.

Indirectly caused: All of AllHere's employees were laid off, resulting in sudden loss of employment and livelihood, and the company entered Chapter 7 bankruptcy.

Inferred additional harm: Other school districts with smaller contracts likely suffered unquantified financial losses due to the sudden termination of services.

Privacy

Reported: The reports describe potential privacy violations regarding the handling of student records by the chatbot.

Directly caused: A former software director alleged that student records and PII were unnecessarily shared with third-party companies and processed on overseas servers.

Inferred additional harm: It is likely that the personal data of thousands of students who interacted with the platform was exposed to security vulnerabilities due to non-compliant processing practices.

People affected

• Occurrences reported: 1
• People reportedly harmed: 50
• People reportedly exposed: 500000

Potential causes

Management

• Fraudulent Financial Inflation: CEO lied about AI revenue and contracts to secure venture capital.
• Embezzlement of Startup Funds: CEO diverted investor cash for personal use, accelerating bankruptcy.
• Overpromising AI Capabilities: Startup management was pressured to overpromise and underdeliver.

Technology

• Unviable AI Chatbot Complexity: Ambitious AI chatbot was too complex for a small text-messaging startup.
• Loss of Human Moderators: The AI required human-in-the-loop oversight which ceased after furloughs.
• Lack of AI Tech Support: Company collapse left the AI without necessary maintenance and training.

Data Inputs

• Insecure Student Data Routing: Student records and chatbot prompts were processed via overseas servers.
• Unauthorised Data Sharing: Prompts with personal student info were shared with third-party companies.

Human Factors

• Lack of Client Vetting Expertise: School districts lacked the technical expertise to vet the AI they bought.
• Susceptibility to AI Hype: District leadership fell for AI hype without verifying product viability.

Process and Methods

• Ignored Security Warnings: Whistleblower alerts about student data privacy violations were ignored.
• Inadequate Financial Auditing: Investors and clients failed to audit startup financials before contracting.

Regulatory Environment

• Weak AI Privacy Protections: Privacy policies failed to protect student data shared with AI contractors.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide clear, detailed, and consistent information from official legal indictments and major news outlets regarding the fraud, the startup's collapse, and the impact on school districts. The role of the AI system and the nature of the human-led fraud are explicitly documented.
• Ambiguities identified: The exact number of laid-off employees is not specified, and the extent of the alleged data privacy breach remains under investigation.
• Alternative interpretations: The incident could be viewed purely as a corporate financial crime rather than an AI safety failure, though the collapse directly terminated an active public-school AI deployment.

The collapse of AI educational startup AllHere due to a multi-million dollar fraud scheme by its founder is primarily a domestic financial crime. The incident has negligible national security implications, with only minor data privacy concerns regarding student records being processed on overseas servers.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Joanna Smith-Griffin

Threat characteristics

• Imminence: Long-term. The company has collapsed and the founder has been arrested, meaning any ongoing threat relates to long-term regulatory and data governance concerns.
• Autonomy: Human-supervised. The AI chatbot operated autonomously to interact with students but required human moderators to oversee the system.
• Novelty: Established threat. Financial fraud, corporate collapse, and data privacy vulnerabilities are well-established threats with numerous historical precedents.

Impact by dimension

• Physical security: Negligible. No physical threats, kinetic attacks, or critical infrastructure impacts were associated with this educational chatbot or the company's financial collapse.
• Information security: Negligible. No intelligence compromise or systematic information warfare occurred. Allegations of student data on overseas servers do not involve classified intelligence.
• Sovereignty: Negligible. The incident involved school district contracts and corporate fraud, which do not threaten core state authority or federal government operations.
• Economic security: Negligible. While there was a multi-million dollar fraud scheme, it represents localized corporate crime rather than a strategic threat to national economic or technological security.
• Societal stability: Minor. The incident involved potential privacy violations of student records and PII processed on overseas servers, representing a minor impact on privacy protections.