Provenance record

Tasmanian justice department review under way after AI and fake citation used in murderer’s parole decision

The Guardian: Artificial Intelligence (tier 1, news) 2026-09-19T06:21:51.000Z Original ↗

Discourse valence

-65

Strongly adverse

confidence 82% · 5 items · range -100 to -40

-100

-50

+50

+100

-65

Adverse readingFavourable reading

Consensus of 5 models from different labs. Spread 60 points, agreement low.

Governance and regulation Synthetic media and manipulation Concentration of power

Excerpt as ingested

Parole condition of convicted killer Susan Neill-Fraser deemed invalid after board cited non-existent case law Get our breaking news email , free app or daily news podcast A review of potential artificial intelligence use in parole decisions will be undertaken after a “concerning” error in the case of a high-profile murderer. Tasmania’s justice department on Friday evening confirmed the review was occurring, four days after a parole condition of convicted killer Susan Neill-Fraser was deemed invalid. Continue reading...

Every model that read this

Model	Provider	Stage	Score	Conf.	Latency	Prompt	When
Llama 3.3 70B	Meta	analysis	-75	80%	3283ms	v1.0.0 / m1.0.1	2026-09-19 09:17
GPT-4.1 mini	OpenAI	consensus	-40	80%	3317ms	v1.0.0 / m1.0.1	2026-09-19 09:18
Claude Sonnet 5	Anthropic	consensus	-45	70%	6369ms	v1.0.0 / m1.0.1	2026-09-19 09:18
Gemini 2.5 Flash	Google	consensus	-100	90%	1622ms	v1.0.0 / m1.0.1	2026-09-19 09:18
Mistral Small 3.1 24B	Mistral AI	consensus	-65	90%	6718ms	v1.0.0 / m1.0.1	2026-09-19 09:18

Llama 3.3 70B · reading

AI error led to invalid parole condition

evidence: reported horizon: n/a

GPT-4.1 mini · reading

The use of AI led to the citation of a fake legal case, resulting in an invalid parole condition for a convicted murderer. This error indicates risk in relying on AI for critical judicial decisions without adequate oversight, which could undermine justice and public trust. However, the discovery and review process suggests potential to improve safeguards.

evidence: reported horizon: near capability 40 societal 50 existential 5 economic 10

Claude Sonnet 5 · reading

A fabricated AI citation directly affected a real parole decision, showing tangible harm from unchecked AI use in a high-stakes legal context. This is a concrete, reported failure rather than speculation, though the scope beyond this single case is unclear.

evidence: reported horizon: n/a

Gemini 2.5 Flash · reading

An AI system generated a fake legal citation, leading to an invalid parole condition for a convicted murderer. This directly undermines the justice system's integrity and accuracy, with severe implications for fairness and public trust in critical legal processes.

evidence: reported horizon: n/a

Mistral Small 3.1 24B · reading

AI was used in a legal decision with a significant error. This indicates a risk of AI causing harm in critical decision-making processes. The review suggests a recognition of this risk.

evidence: primary horizon: n/a existential 0

Evidence extracted

Tasmania's justice department uses AI in parole decisions
AI cited non-existent case law in a murderer's parole decision

The chain

SOURCE     The Guardian: Artificial Intelligence (tier 1)
   ↓
DOCUMENT   a27b0e3f-8637-4352-a73e-e0ca86747954
           https://theguardian.com/australia-news/2026/sep/19/tasmania-justice-department-review-ai-artificial-intelligence-fake-citation-murderer-parole-decision-ntwnfb
   ↓
EVIDENCE   2 extracted excerpts
   ↓
MODEL RUN  5 runs, methodology 1.0.1
   ↓
SCORE      -65  (Strongly adverse)
   ↓
CONFIDENCE 82%