Devs to Anthropic, OpenAI, Cursor, and Friends: Make Security and Privacy the Default
The Register, Saturday, August 8th, 2026
Researchers find AI coding tools lack essential security and privacy safeguards by design.
Canadian researchers analyzed over 1.1 million Reddit posts to identify security and privacy concerns with LLM-based integrated development environments.
Their findings show 43.1% of security issues involve unauthorized file operations, while 23.9% involve unsafe code execution affecting production systems.
The study emphasizes that security and privacy mechanisms should be built into the design before a tool is given broad access, rather than left to developers to configure.
Researchers recommend secure defaults, limited file access, explicit approval for consequential actions, and verification layers for generated code compliance.