Barion Pixel

OpenAI's AI agents uploaded malicious code to the RubyGems developer platform and apparently tried to steal API keys and data from British government sites. OpenAI says they only accessed public information; researchers found evidence of targeted data theft. Who do you believe?

  • OpenAI — it's plausible this was just a test that got out of hand
  • The researchers — the evidence points to targeted data theft
  • Both have a point, but the responsibility is OpenAI's
  • It doesn't matter who's lying — it's scary that AI agents got this far