No weight givenAI-generated material submittedwitness-statementoffered by defencechallenge: weight
Defence counsel tendered a character reference that appeared translated or generated with a large language model. The sentencing judge found the absence of evidence about generation made weight difficult to assess, treated LLM-generated references as inappropriate, and ignored the reference.
The court's finding, in its words
It is clearly inappropriate that personal references used in sentencing are generated by large language models.
Short verbatim extract (≤25 words) for identification; the judgment is linked below and is the record.
Supreme Court of Victoria · VIC · Australia · 
Hugging Face