Latest articles

RSS
  1. Technology

    Brazilian court sanctions hidden AI instructions in a legal filing

    Outside documents are material to review, not commands to follow; one detected attack does not establish general protection.

    • AI security
    • Prompt injection
  2. Technology

    Epoch AI studies GLM 5.3 attacks across Firefox's patch gap

    Public fixes can reveal attack clues before delivery; the study does not establish the safety of the latest version.

    • AI security
    • Patching
  3. Technology

    Browser-fingerprinting code uses hostile text to disrupt AI analysis

    Alarmingly named functions and injected strings produced reactions unrelated to the code's actual behavior.

    • AI security
    • Prompt injection
  4. Technology

    Korean police refer suspects in AI-assisted data theft case

    Reported record totals differ from the initial charges and from unique victim counts.

    • AI security
    • Data breaches
    • Incident response
  5. Technology

    Fortinet adds inline AI interaction inspection for FortiGate

    Cloud-delivered inspection and self-hosted gateways have different data boundaries.

    • AI security
    • MCP
    • Gateways
  6. Technology

    Cathedral's Anthropic access report centers on third-party business software

    Reported limited access does not establish exposure of customer conversations or model weights.

    • AI security
    • SaaS
    • Access controls

You’ve reached the end.

Looking for something specific? Search this section
Search

What are you curious about?

Find stories by title, topic, source or anything in the article.