positive.security/blog/auto-gpt-rce

Preview meta tags from the positive.security website.

Linked Hostnames

2

Thumbnail

Search Engine Appearance

Google

https://positive.security/blog/auto-gpt-rce

Hacking Auto-GPT and escaping its docker container | Positive Security

We leverage indirect prompt injection to trick Auto-GPT (GPT-4) into executing arbitrary code when it is asked to perform a seemingly harmless task such as text summarization on a malicious website, and discovered vulnerabilities that allow escaping its sandboxed execution environment.



Bing

Hacking Auto-GPT and escaping its docker container | Positive Security

https://positive.security/blog/auto-gpt-rce

We leverage indirect prompt injection to trick Auto-GPT (GPT-4) into executing arbitrary code when it is asked to perform a seemingly harmless task such as text summarization on a malicious website, and discovered vulnerabilities that allow escaping its sandboxed execution environment.



DuckDuckGo

https://positive.security/blog/auto-gpt-rce

Hacking Auto-GPT and escaping its docker container | Positive Security

We leverage indirect prompt injection to trick Auto-GPT (GPT-4) into executing arbitrary code when it is asked to perform a seemingly harmless task such as text summarization on a malicious website, and discovered vulnerabilities that allow escaping its sandboxed execution environment.

  • General Meta Tags

    7
    • title
      Hacking Auto-GPT and escaping its docker container | Positive Security
    • charset
      utf-8
    • description
      We leverage indirect prompt injection to trick Auto-GPT (GPT-4) into executing arbitrary code when it is asked to perform a seemingly harmless task such as text summarization on a malicious website, and discovered vulnerabilities that allow escaping its sandboxed execution environment.
    • twitter:title
      Hacking Auto-GPT and escaping its docker container | Positive Security
    • twitter:description
      We leverage indirect prompt injection to trick Auto-GPT (GPT-4) into executing arbitrary code when it is asked to perform a seemingly harmless task such as text summarization on a malicious website, and discovered vulnerabilities that allow escaping its sandboxed execution environment.

  • Open Graph Meta Tags

    4
    • og:title
      Hacking Auto-GPT and escaping its docker container | Positive Security
    • og:description
      We leverage indirect prompt injection to trick Auto-GPT (GPT-4) into executing arbitrary code when it is asked to perform a seemingly harmless task such as text summarization on a malicious website, and discovered vulnerabilities that allow escaping its sandboxed execution environment.
    • og:image
      https://cdn.prod.website-files.com/5f6498c074436c349716e747/649c274942ea5052b44ec83e_Auto-GPT-RCE-exploit-path.png
    • og:type
      website

  • Twitter Meta Tags

    1
    • twitter:card
      summary_large_image

  • Link Tags

    7
    • alternate
      rss.xml
    • apple-touch-icon
      https://cdn.prod.website-files.com/5f6498c074436c50c016e745/5f7dd71edeeceb5d47162386_256_256.png
    • preconnect
      https://fonts.googleapis.com
    • preconnect
      https://fonts.gstatic.com
    • shortcut icon
      https://cdn.prod.website-files.com/5f6498c074436c50c016e745/5f7ddb13deeceb266b162f8d_favicon-32x32_white.png

Emails

1

Links

8