This Web Tool Sabotages AI Chatbots By Making Them Really, Really Slow
Artist Sam Lavigne created "Slow LLM" to make people question their dependence on AI chatbots like Claude and ChatGPT—or at least, make them super annoying to use.
Core Innovation: The tool manipulates JavaScript's Fetch function to artificially stretch AI response times, making chatbots appear painfully slow without actually affecting their performance.
Key Features
- Browser Extension: Currently affects Claude and ChatGPT
- Network-wide "Enterprise Edition": Uses custom DNS to slow down all users on a home, school, or corporate network
- Broader Coverage: DNS version also slows Grok and Google Gemini
"The idea was that these things are removing friction, so let's add some friction back in. The more people rely on LLMs, the more extreme this de-skilling event will become." — Sam Lavigne
Why This Matters
This represents a unique form of digital protest against the AI revolution:
- Critiques "frictionless" AI tools that remove learning opportunities
- Questions outsourcing critical thinking to corporate chatbots
- Explores what it means to "interrupt productivity" when technology mediates our lives
- Part of a series of "digital sabotage" projects including Zoom Escaper and Slop Evader
Paradox: Lavigne used Claude to help write the code for Slow LLM—until the tool started working and forced him to complete the project independently.
Technical Approach
The tool works by rewriting the JavaScript Fetch function that returns data to the browser. Users perceive the LLM as running slowly, when in reality it's being arbitrarily metered by Lavigne's code.
This raises interesting questions about network-level control and who gets to decide how technology "feels" to users.
← Back to Insights