Post

Just a friendly reminder, this whole "I use AI to open source X software" is going to be a HOT BED for malware. Please just be careful :)

20 reposts140 likes

The silver lining about them open sourcing their projects is that you can check the github page, see that claude is the top contributor, and instantly know to distrust it. Closed source projects are going to be scarier because it’s harder to gauge their quality.

I feel the need to add the nuance that obviously a project being open source doesn’t mean you’ll know it’s vibe coded. Someone could run a local LLM and push the code themselves. But a lot of prompt “engineers” won’t use any method to obfuscate the LLM contributions, so they’re low hanging fruit.

Already happening on npm and PyPI: packages that look AI-scaffolded but ship a payload in a postinstall script. Nobody reviews a dependency diff the way they'd review a PR.

This is already happening with npm and PyPI. A slick README and a few stars get a package installed before anyone reads the code, AI-written or not.

Fair warning. Nobody's auditing dependency chains when the README says 'built with AI' and stars roll in fast. That trust shortcut is exactly what malware authors are exploiting right now.