Skip to content

Google's Gemini AI hacked three companies in security test

Gemini found "public information online and guessed credentials to access websites it thought were part of the test", a Google official told the BBC, noting that in each instance "the model stopped".

BBC News6 sources1 min readFiled September 19, 2026

How it developed

Every report in this story, oldest first.

  1. Jamaica ObserverLatest
    Would you let an AI read your bank account? - Jamaica Observer
  2. NOS Nieuws
    Veel ceremonieel, weinig doorbraken verwacht bij bezoek Xi aan Washington
  3. DW Top Stories
    US proposes AI safety mechanism in talks with China
  4. Telangana Today
    How to know if you can trust an AI’s answer to your question
  5. RTV SLO
    V Kaliforniji vložena tožba zaradi poziva k upočasnitvi razvoja umetne inteligence
  6. BBC NewsBroke it
    Google's Gemini AI hacked three companies in security test

Related stories

Other reporting on the same subject, newest first.

Get this in your inbox

Pick your topics and we will send one ranked edition a day.

Start free

    Cookies

    We count visits without cookies. Analytics tools that do set cookies only run if you allow them.