Plaintiff sanctioned for hiding invisible AI prompts in court filings
The Decoder · rss · 2026-08-15
A plaintiff in Connecticut embedded invisible prompt injections (3-point white text on white background) in court filings to manipulate a potential AI review system. Although the court stated it does not use AI for review, Judge Spader compared the act to tampering with a jury and revoked the plaintiff's electronic filing privileges. The court ruled that the intent alone justified the sanctions.
More from Safety
- Abliteration benchmarks become targets for regulating open-weight models — StandardLovers · 2026-08-15
- DeepMind Non-Compete Agreements Hinder UK AI Startups — NandoDF · 2026-08-15
- Anthropic reveals Claude text watermarking details, built into latest models — 新智元 · 2026-08-15
- Amazon allows using Twitch content to train AI unless users opt out — Wired AI · 2026-08-15
- tl;dv data isolation flaw exposes 180k meeting records — emmanuelvivier · 2026-08-15
- EU AI Act transparency rules take effect: chatbots and deepfakes require clear labels — emmanuelvivier · 2026-08-15