Alaska Education Department Reportedly Published Policy Featuring Erroneous AI-Generated Citations

In October 2024, Alaska’s Education Commissioner Deena Bishop allegedly used generative AI to draft a school cellphone policy, resulting in reportedly fabricated citations of non-existent studies. The document, which did not disclose AI use, was shared with the state Board of Education, and despite reported corrections, AI-generated errors remained in the final resolution.

Risk classification

• Primary risk domain: 3 Misinformation
• Primary risk subdomain: 3.1 False or misleading information

The generative AI system fabricated academic citations that were published in an official state policy document, spreading false information to policymakers and the public.

Additional risk subdomains

• 5.1 Overreliance and unsafe use: State officials relied on the AI system to draft policy and generate citations without verifying the accuracy of the outputs.
• 7.3 Lack of capability or robustness: The AI system failed to perform reliably by hallucinating non-existent sources, demonstrating a technical limitation in critical applications.

Causal factors

• Entity: AI
• Intent: Unintentional
• Timing: Post-deployment

The risk was caused by the generative AI system hallucinating non-existent academic citations during the drafting process, which was an unexpected outcome of using the tool.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The AI system used was a generative AI platform for content generation, which is subject to transparency obligations under the EU AI Act to disclose that the content was AI-generated.

AI system and alleged parties

• AI system: generative AI platform
• AI purpose: Writing Assistant; Content Search
• Behaviour type: Assistant
• Alleged developer: Unknown large language model developers
• Alleged deployer: Deena Bishop, Alaska Department of Education and Early Development
• Alleged harmed parties: Alaskan students, Alaskan parents of students, Alaskan education system, Alaskan education policymakers, Alaska Board of Education members

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. 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 Negligible, indirect Negligible
• 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 Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Minor
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Epistemic

Reported: Yes, the report explicitly describes the generation of fabricated academic citations by the AI system.

Directly caused: The AI system generated four false citations of non-existent studies in scientific journals, which were included in a draft state resolution.

Indirectly caused: The publication of these fake citations on the department's website misled the public and board members, polluting the information ecosystem.

Inferred additional harm: The continued presence of 'vestiges' of AI hallucinations in the updated document suggests that unverified AI outputs can permanently contaminate official records.

People affected

• Occurrences reported: 1
• People reportedly exposed: 10

Potential causes

Management

• Lack of Accountability: Management did not disclose AI use in the conception of the document.
• Inadequate Risk Assessment: Failure to assess the impact of AI misinformation on public trust.

Technology

• AI Hallucination: The model generated fake citations that resembled real academic studies.
• Lack of Source Verification: AI lacks built-in mechanisms to verify truthfulness of generated URLs.

Data Inputs

• Insufficient Training Data: The AI model lacked accurate real-time data to retrieve genuine sources.

Human Factors

• Low AI Literacy: Officials treated generative AI as a reliable search tool rather than LLM.
• Overreliance on AI Output: Commissioner trusted the AI's generated citations without manual checking.

Process and Methods

• Inadequate Review Process: The department failed to verify citations before publishing the draft.
• Use of AI Placeholders: Drafting process relied on unverified AI text as temporary filler.

Regulatory Environment

• Absence of AI Use Policy: No state regulations exist governing AI use in official policy drafting.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear, detailed, and uncontested facts about the state official's use of generative AI, the specific fake citations generated, and the subsequent corrections and remaining errors. The role of the AI as a source of hallucinated information is explicitly confirmed by both the department spokesperson and an external AI expert.
• Ambiguities identified: The specific generative AI platform used (e.g., ChatGPT, Gemini) is not named, referred to only as 'a generative AI platform'.
• Alternative interpretations: None. The facts of the AI's hallucination and its use by the commissioner are clearly established and admitted by the department.

The Alaska Department of Education used generative AI to draft a school policy resolution containing fabricated citations. While it represents a minor compromise of state-level policy-drafting integrity, it poses negligible national security risks and was managed through standard administrative corrections.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Alaska Department of Education and Early Development

Threat characteristics

• Imminence: Long-term. The immediate incident is resolved, but it highlights an ongoing, long-term strategic concern regarding the unverified use of AI in public administration.
• Autonomy: Human-controlled. The AI system was used strictly as a drafting assistant, with final policy decisions and oversight remaining with human officials.
• Novelty: Established threat. AI hallucinations and the failure to verify generative AI outputs are well-documented and established technical risks.

Impact by dimension

• Physical security: Negligible. No physical threats, kinetic targeting, or critical infrastructure disruptions were reported in this incident.
• Information security: Negligible. The incident involved unintentional AI hallucinations in a local policy document, not a coordinated information warfare or intelligence operation.
• Sovereignty: Minor. Minor impact on state-level administrative and policy-drafting processes due to unverified AI-generated content, manageable within standard procedural reviews.
• Economic security: Negligible. No impact on national economic stability, financial systems, or critical technological infrastructure.
• Societal stability: Negligible. No threats to societal stability, civil liberties, or human rights were observed in this administrative incident.