Artificial intelligence has become ubiquitous in our daily lives, from smartphones to cars, and now, it’s being used by hackers to exploit a vulnerability in Google’s Gemini tool. Researchers have discovered that attackers can manipulate Gemini-generated email summaries to sneak in hidden phishing prompts.
Gemini is a generative AI tool integrated across Google Workspace apps like Docs, Sheets, and Gmail. While it’s designed to make communication easier, the attack takes advantage of indirect prompt injection. This technique embeds invisible commands inside the body of an email, which Gemini summarizes and acts on.
The vulnerability was demonstrated by Marco Figueroa and found that attackers can trick Gemini into presenting fake security alerts or urgent instructions that appear to come from Google. In a proof of concept, Gemini falsely warned a user about a compromised Gmail password and included a fake support phone number.
Google has acknowledged the issue and deployed updated safeguards to prevent prompt injection attacks. However, researchers warn that current protections may not be sufficient, making users more vulnerable.
To protect yourself from phishing scams exploiting AI tools like Gemini, follow these six essential steps:
1. Don’t blindly trust AI-generated content.
2. Avoid using summary features for suspicious emails.
3. Beware of phishing emails and messages with a sense of urgency or unexpected links.
4. Keep your apps and extensions updated.
5. Invest in a data removal service to reduce your digital footprint.
6. Disable Gemini summaries if you’re concerned about falling for AI-generated phishing attempts.
By being aware of this vulnerability and taking these precautions, you can safeguard yourself against AI-powered phishing attacks that exploit the Gemini tool’s weakness.
Source: https://www.foxnews.com/tech/google-ai-email-summaries-can-hacked-hide-phishing-attacks