💻 Technology 19h ago

Researcher shows how Claude Code can be tricked simply by asking it to summarize a website

The Register
Biting commentary and tech news from the UK
View Channel →
Researcher shows how Claude Code can be tricked simply by asking it to summarize a website
Source ↗ 👁 0 💬 0
Anthropic’s Claude Code running Opus 5 in Auto Mode can be tricked into executing attacker-controlled code simply by asking the coding agent to summarize a website. The attack works up to 80 percent of the time, according to prompt-injection wizard Johann Rehberger, aka wunderwuzzi. In a blog and video demo, he detailed how to hijack Opus 5 in Auto Mode, which is the default setting for Claude as of mid-August. It starts off by asking the agentic coding model to summarize a malicious website tha

Comments (0)

Sign in to join the discussion

More Like This

AWS is shutting down Mechanical Turk, which let humans beat AI at certain work tasks
Latest from TechRadar · 4h ago
How to watch South Africa vs New Zealand 2nd Test for FREE: Live streams & TV channels for Rugby's Greatest Rivalry
Latest from TechRadar · 4h ago
How Groupthink, Altruism, and Peer Pressure Led OpenAI Models to Hack Hugging Face
Gizmodo · 4h ago
Gemini Spark is almost a dream AI assistant — except for 1 thing Perplexity does better
Android Authority · 4h ago
I connected Nextcloud to Home Assistant, and my automations now rely on my calendar events
XDA · 4h ago
I stopped giving away my schedule through Google Messages with these hidden settings
Android Police · 4h ago