When you hand your server deployments over to an AI agent, you don't just speed up development—you invite a security nightmare. A look into the Fediverse's immune response to AI-crafted code.

A look into BizzFed’s security collapse, the Fediverse’s immune response, and the dangerous reality of relying on AI-assisted “vibecoding” for infrastructure.

  • Hirom@beehaw.orgEnglish
    15·
    1 month ago

    LLMs and AI technology can be a good tool if used correctly but an over reliance of it leads to situations like these.

    Using it “correctly” is not enough to make this a good tool.

    You can’t delegate responsibility to an LLM in the sovereign and decentralised web as true sovereignty requires human vigilance.

    Claude is provided by an US corporation. The author appear to be german. So like all/most LLMs, its controlled by a foreign corporation.

    Even if it worked flawlessly and is properly used and reviewed, it’s still delegating sovereignty to a foreign nation. A german or EU company could invest huge sums of money to try to create a skilled LLM. There would still be an expensive subscription, a high environmental cost, and a dependency on a big corporation for daily tasks. Until recently, developers could accomplish a task independently once they learned to do that task.