Tasmania audits parole decisions citing fake AI case law

By Diana Trent · 2026-09-19

What the court found

Justice Stephen Estcourt of the Supreme Court of Tasmania struck down a parole condition on 15 September 2026. The condition had barred Susan Neill-Fraser from communicating with any media outlet to claim her innocence or wrongful conviction.^[1] His Honour held the condition was both procedurally unfair and legally unreasonable. The board had relied on a document citing case law that did not exist. The fictitious citations bore the hallmarks of generative AI output.

The media-ban condition had been added to Neill-Fraser's parole in December 2024, prohibiting her from communicating directly or indirectly with any media outlet to assert her alleged innocence and wrongful conviction.^[2] Neill-Fraser was released on parole on 4 October 2022 after serving thirteen years of a twenty-three-year sentence for the 2009 murder of Bob Chappell.^[3] The Human Rights Law Centre filed proceedings on her behalf in April 2025 challenging the condition.

Sarah Schwartz, Legal Director of the Human Rights Law Centre, said the ruling confirmed what the challenge had always argued. "Freedom of speech and political communication are core democratic rights that belong to everyone, including and especially people who have been in prison, whose voices are essential to a just and democratic society," Schwartz said.^[4]

The government's response

On 16 September 2026, Attorney-General Guy Barnett wrote to the chair of the Parole Board demanding a written explanation of its professional standards, describing the use of AI-generated case law as "unacceptable".^[5] Three days later, the Justice Department announced an independent review into the extent to which AI had informed past Parole Board decisions.^[6]

Barnett said: "The production of inaccurate information by a public authority is unacceptable; I have asked the Parole Board for an explanation of its professional standards."^[5] The scope of the review extends to all decisions where AI use can be identified, not only those involving Neill-Fraser.

Broader implications for Australian tribunals

The case appears to be the first in Australia where a hallucinated AI submission has directly voided a formal tribunal decision. Cassy O'Connor, the Greens justice spokesperson, said the incident "raises bigger questions about the use of AI in a manner that could lead to justice being denied to Tasmanians who come before the courts."^[7]

Justice Estcourt's finding that reliance on fictitious case law deprived Neill-Fraser of procedural fairness gives the ruling weight beyond its facts. It establishes that an unchecked AI-generated legal submission can breach natural justice badly enough to void a decision.^[1] That reading will be studied by tribunals and statutory boards across Australia that have begun integrating generative AI into document preparation without published protocols.

The Justice Department has not set a public deadline for completion of the audit, and the Parole Board's written response to Barnett's letter had not been released as of the date of this report.

Diana Trent

Diana Trent writes about regulation, competition and the law as it meets technology. She reads the judgments and the regulator filings that most people skip, and finds the story in them.