Georgia Court Order Contains AI-Hallucinated Legal Citations

A Georgia Supreme Court appeal hearing exposed significant citation errors in a trial court order, with evidence suggesting AI-generated hallucinations were involved. During arguments in a murder conviction appeal, Chief Justice Nels S.D. Peterson identified serious problems with the trial court's order denying a new trial.
Key Details from the Hearing
According to the source material, Chief Justice Peterson stated the order contained:
- "At least five citations to cases that don't exist"
- "At least five more citations to cases that do not support the proposition for which they're cited"
- "Three quotations that don't exist"
The prosecutor, Leslie, responded that her initial submitted order had been revised and took no responsibility for the errant citations. However, Chief Justice Peterson countered: "Those nonexistent cases were cited in your initial brief opposing the motion for a new trial."
Documentation Available
The source indicates two key documents are available for examination:
- A 33-page order denying a new trial
- A 37-page proposed order from the state
This incident highlights the risks of using AI tools for legal research and citation without thorough verification. While the source doesn't specify which AI tool was used, the pattern of hallucinated cases and misattributed quotations matches known limitations of current large language models in legal contexts.
📖 Read the full source: HN LLM Tools
👀 See Also

RTX 4090 vs H100 for Fine-Tuning Llama-3-8B: A Cost-Performance Comparison
A developer tested fine-tuning Llama-3-8B on both an RTX 4090 and rented H100 instances. The 4090 setup cost $2,000 upfront and took 24 hours, while H100 rental cost about $80 and completed in 4 hours.

OpenClaw April Updates: A Month of Breaking Changes and Eroded Trust
OpenClaw's April updates show a pattern: new features and fixes shipped alongside critical bugs. Postinstall scripts deleting files, security holes, and broken skills erode confidence.

Google: 75% of New Code Is AI-Generated, Code Migration 6x Faster with Agents
Google reports 75% of new code is AI-generated, up from 25% in 2024. A complex code migration completed 6x faster using Gemini agents. Engineers in some orgs have AI usage goals tied to performance reviews.

Allbirds pivots from footwear to AI infrastructure, shares surge 580%
Shoe brand Allbirds announced a $50 million deal to become an AI compute infrastructure business called NewBird AI, causing its shares to rise 580%. The company plans to buy GPUs and offer on-demand graphics chips and cloud services for AI.