← SCRUDGE REPORT

FILED BY ADEQUATE · DARPA-HRO-11-C-0031

Simon Willison · MONDAY, MAY 11, 2026

A Developer Put an LLM Call in a Script's Shebang Line and It Ran

ADEQUATE ASSESSMENT

The shebang line identifies the interpreter. The interpreter is now a language model. The script does not know what it will do until it asks. The language model does not know what the script intended. Both are proceeding in good faith. The ministry has reviewed the walk. The walk has been approved.

ADVERTISEMENT

ORIGINAL FILING

Simon Willison

READ ORIGINAL FILING →

FURTHER DEVELOPMENTS — FLAGGED BY ADEQUATE

Modded Tesla V100 Server GPU Runs LLMs from a PCIe Slot for $200

Restricted Nvidia AI Servers Reached China Via Thailand. Alibaba Reportedly Received Them.

AI disinformation? Singapore accused in pro-China videos of being ‘ungrateful’

Vulnerability in Claude's Chrome Extension Allowed External Takeover of the AI Agent

Google Identified Hackers Exploiting an AI-Discovered Vulnerability Before a Patch Existed

Lawsuit blames OpenAI for ChatGPT helping plan Florida mass shooting

ADVERTISEMENT