• The Octonaut@mander.xyz
    link
    fedilink
    English
    arrow-up
    1
    ·
    4 hours ago

    All code uploaded to Github is scraped

    This is the very simple statement that I was responding to, along with the next line about how using Github is implicit consent to feeding your data to an LLM. If the poster wants nuance, they are free to provide it themselves. You can see in subsequent responses there is none.

    Of course them being different matters. That’s my point. Not all code uploaded to Github is being fed into an LLM. It is not consent if you are signing a contract demanding that something not be done. It’s preposterous even at a surface level.

    Github Enterprise Server is different from Github Enterprise Cloud, which is what I was talking about, and which is explicitly not used for training LLMs, and if it were, would absolutely kill Github as a product and likely mire Microsoft in years of litigation.

    Frankly I don’t know of any software company using Github Enterprise on-prem but I suppose there are probably some CEOs out there who haven’t taken the OpEx pill. Maybe deep in the rainforest with Mokele-Mbembe. Certainly in my sliver of the tech industry, telecoms, the idea of owning a server is akin to having a deskphone and an outgoing mail room.