AI Reading List

Daily curated list of news, articles, and other interesting posts related to AI.

2025-11-24

• A free tool that stuns LLMs with thousands of invisible Unicode characters (gibberifier.com)

  Overview of a tool that blocks AI models from reading text by embedding invisible Unicode characters, making the content unreadable for language models but still legible for humans. Tested across several popular AI systems, the approach often causes confusion, crashes, or failure to process the altered text.

• The Bitter Lesson of LLM Extensions (www.sawyerhood.com)

  Overview of how LLM customization has evolved, from early ChatGPT plugins and manual prompts to more advanced methods like memory, repository-level instructions, and tool integrations. Newer systems aim to balance power with ease of use, highlighting ongoing tension between feature complexity and user-friendly design.