AI Medical Scribe Reportedly Added False Psychedelic-Mushroom Use Claim to Australian Patient Rebecca Green's Medical Record

Australian patient Rebecca Green reportedly discovered that a post-operative letter sent by her urologist to her GP falsely said she micro-dosed psychedelic mushrooms and suggested this could explain prior kidney bleeding. Green had reportedly consented to an AI scribe during her first urology appointment and feared the false record could affect her workers' compensation case. Her urologist reportedly apologized, corrected the letter, and said the error appeared to arise during dictation or transcription.

Risk classification

• Primary risk domain: 7 AI system safety, failures, & limitations
• Primary risk subdomain: 7.3 Lack of capability or robustness

The primary failure lies within the AI system's lack of capability and robustness, specifically its propensity to hallucinate false symptoms, conditions, and drug use during transcription.

Additional risk subdomains

• 5.1 Overreliance and unsafe use: The doctor overrelied on the AI scribe by failing to check the output before sending it, illustrating the human tendency to stop verifying AI outputs over time.
• 3.1 False or misleading information: The AI system generated discrete false and misleading information regarding the patient's drug habits, which polluted her official medical record.

Causal factors

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

The incident was caused by an AI scribe hallucinating false medical information during post-deployment operation, which was an unexpected and unintentional outcome of its transcription goal.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The AI system is used in healthcare as a clinical documentation tool, which has significant implications for patient safety and falls under the category of AI used in healthcare.

AI system and alleged parties

• AI system: AI medical scribe
• AI purpose: Writing Assistant; Health AI Assistant
• Behaviour type: Assistant
• Alleged developer: AI transcription technology developers
• Alleged deployer: Medical personnel, Doctors, AI transcription tool users
• Alleged harmed parties: Rebecca Green, Privacy, Patients, Epistemic integrity

Harm severity

Severity ratings are based on this scale.

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

Psychological

Reported: The report explicitly describes psychological distress and anxiety caused to the patient.

Indirectly caused: The patient, [PERSON_002], was 'gobsmacked and in tears' and suffered significant stress and fear upon discovering the false drug allegation on her medical record.

Epistemic

Reported: The report explicitly describes epistemic harm through the fabrication of medical facts.

Directly caused: The AI scribe fabricated the false claim that the patient micro-dosed psychedelic mushrooms.

Indirectly caused: The fabricated claim was integrated into the patient's official medical record and sent to her GP, corrupting the accuracy of her clinical history.

Inferred additional harm: Given that 40 percent of GPs use these tools and often stop checking them, it is highly likely that many other patients have uncorrected, fabricated errors on their medical records.

People affected

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

Potential causes

Management

• Inadequate Practice Governance: The medical practice lacked policies ensuring clinicians checked AI outputs.
• Prioritizing Speed Over Safety: The practice adopted AI to reduce admin work without maintaining safety checks.

Technology

• AI Hallucination: The AI scribe made up a false claim that the patient took illegal drugs.
• Algorithmic Inaccuracy: The AI tool confused body sides and clinical details like breast cancer location.
• Unreliable Transcription: The transcription process introduced errors that sounded plausible but were false.

Data Inputs

• Acoustic Transcription Errors: The tool misheard spoken words during the doctor-patient consultation.
• Lack of Input Verification: The software did not verify transcribed facts against clinical evidence.

Human Factors

• Automation Bias: The doctor stopped checking the AI output after initial accurate experiences.
• Coerced Patient Consent: The patient agreed to AI use to avoid being perceived as a difficult patient.
• Inadequate Physician Review: The urologist failed to review the post-operative letter before sending it.

Process and Methods

• Lack of Output Verification Process: There was no mandatory verification process for AI outputs before clinical use.
• Inconsistent Consent Methods: Consent was obtained through passive signs or forced online booking systems.
• Deficient Error Correction Flow: Patients faced significant frustration finding and correcting medical errors.

Regulatory Environment

• Exemption from Oversight: Scribes claiming only to transcribe are exempt from TGA medical device rules.
• Lack of Approved Software Lists: Australia lacked a centralized list of tested and approved AI medical tools.
• Unregulated Medical Scribes: The AI scribes market operated without mandatory safety testing standards.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear, firsthand accounts of the patient's experience, quotes from medical professionals, and references to a watchdog report by Digital Rights Watch, making the details highly reliable.
• Ambiguities identified: The specific brand or developer of the AI scribe used by the urologist is not named.

An unregulated AI medical transcription tool hallucinated false drug use in an Australian patient's record due to a doctor's failure to verify the output. While highlighting significant clinical safety, regulatory, and privacy concerns in healthcare AI adoption, the incident poses negligible threat to national security.

• Overall national security impact: Negligible
• Scope: Single nation
• Primary target: Australia
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. Represents an ongoing regulatory and safety concern regarding AI integration in healthcare, rather than an active national security crisis.
• Autonomy: Human-supervised. The AI scribe operates as an assistant where a doctor is expected to review, verify, and sign off on the transcript, though automation bias led to a failure in supervision.
• Novelty: Established threat. AI hallucinations and human overreliance on automated systems are well-documented, established failure modes in natural language processing applications.

Impact by dimension

• Physical security: Negligible. While medical errors can lead to physical harm, this specific incident of transcription hallucination did not result in physical injury, kinetic attacks, or critical infrastructure compromise.
• Information security: Negligible. The incident involves localized medical transcript errors rather than state-sponsored intelligence compromise, classified data theft, or systematic information warfare operations.
• Sovereignty: Negligible. No threat to state authority, territorial control, electoral systems, or core government administrative operations was observed.
• Economic security: Negligible. The incident caused personal concern regarding workers' compensation, but did not threaten national financial systems, critical supply chains, or strategic technological competitive advantage.
• Societal stability: Negligible. The incident caused individual psychological distress and highlighted privacy concerns, but does not represent mass surveillance, systematic discrimination, or large-scale threats to social cohesion.