arstechnica.com/information-technology/2024/01/ai-poisoning-could-turn-open-models-into-destructive-sleeper-agents-says-anthropic/>

Preview meta tags from the arstechnica.com website.

Linked Hostnames

11

Thumbnail

Search Engine Appearance

Google

https://arstechnica.com/information-technology/2024/01/ai-poisoning-could-turn-open-models-into-destructive-sleeper-agents-says-anthropic/%3E

AI poisoning could turn models into destructive “sleeper agents,” says Anthropic

Trained LLMs that seem normal can generate vulnerable code given different triggers.



Bing

AI poisoning could turn models into destructive “sleeper agents,” says Anthropic

https://arstechnica.com/information-technology/2024/01/ai-poisoning-could-turn-open-models-into-destructive-sleeper-agents-says-anthropic/%3E

Trained LLMs that seem normal can generate vulnerable code given different triggers.



DuckDuckGo

https://arstechnica.com/information-technology/2024/01/ai-poisoning-could-turn-open-models-into-destructive-sleeper-agents-says-anthropic/%3E

AI poisoning could turn models into destructive “sleeper agents,” says Anthropic

Trained LLMs that seem normal can generate vulnerable code given different triggers.

  • General Meta Tags

    13
    • title
      AI poisoning could turn models into destructive “sleeper agents,” says Anthropic - Ars Technica
    • charset
      utf-8
    • viewport
      width=device-width, initial-scale=1
    • robots
      max-snippet:-1,max-image-preview:large,max-video-preview:-1
    • description
      Trained LLMs that seem normal can generate vulnerable code given different triggers.

  • Open Graph Meta Tags

    10
    • og:type
      article
    • US country flagog:locale
      en_US
    • og:site_name
      Ars Technica
    • og:title
      AI poisoning could turn models into destructive “sleeper agents,” says Anthropic
    • og:description
      Trained LLMs that seem normal can generate vulnerable code given different triggers.

  • Twitter Meta Tags

    8
    • twitter:card
      summary_large_image
    • twitter:title
      AI poisoning could turn models into destructive “sleeper agents,” says Anthropic
    • twitter:description
      Trained LLMs that seem normal can generate vulnerable code given different triggers.
    • twitter:image
      https://cdn.arstechnica.net/wp-content/uploads/2024/01/AI_sleeper_agent_hero-1152x648.jpg
    • twitter:image:alt
      An illustration of a cyborg "sleeper agent."

  • Link Tags

    11
    • apple-touch-icon
      https://cdn.arstechnica.net/wp-content/uploads/2016/10/cropped-ars-logo-512_480-300x300.png
    • canonical
      https://arstechnica.com/information-technology/2024/01/ai-poisoning-could-turn-open-models-into-destructive-sleeper-agents-says-anthropic/
    • icon
      https://cdn.arstechnica.net/wp-content/uploads/2016/10/cropped-ars-logo-512_480-60x60.png
    • icon
      https://cdn.arstechnica.net/wp-content/uploads/2016/10/cropped-ars-logo-512_480-300x300.png
    • preconnect
      https://c.arstechnica.com

Emails

1

Links

62