A security researcher uncovered a hidden line of code that allowed Anthropic to collect data on Chinese users without detection.

  • AwesomeLowlander@sh.itjust.works
    cake
    link
    fedilink
    English
    arrow-up
    12
    arrow-down
    1
    ·
    3 days ago

    TLDR it was aimed at detecting / blocking competing AI companies trying to copy their model. Ironic given the provenance of most training data, but not entirely malware per se.