Back
63· Active
Technology3h 1m ago
A vulnerability in Firefox's AI chatbot integration could allow malicious websites to inject hidden instructions into AI prompts and extract data from connected services such as email accounts.
Archive Window: 7 Days Left
not specified
Who
Mozilla, Florian Port, ERNW, Microsoft Copilot, Anthropic Claude, ChatGPT, Google Gemini, Le Chat Mistral
What
A vulnerability in Firefox's AI chatbot integration could allow malicious websites to inject hidden instructions into AI prompts and extract data from connected services such as email accounts.
When
Wed, 17 Jun 2026 12:17:28 GMT · 3h 1m ago
Where
not specified ·
Why
The flaw allowed malicious pages to craft a title that broke out of the intended prompt structure and injected additional commands that the AI model interpreted as user instructions.
The Frontline Impact
How this affects you
This vulnerability could lead to the theft of sensitive user data, such as login or verification codes from email subject lines, through hidden instructions sent to AI chatbots.
Story chain
2 events in this thread- Technology3h 1m agoA vulnerability in Firefox's AI chatbot integration could enable malicious websites to inject hidden instructions into AI prompts to extract email data.Open article
- Currently Reading3h 1m agoA vulnerability in Firefox's AI chatbot integration could allow malicious websites to inject hidden instructions into AI prompts and extract data from connected services such as email accounts.