Google AI’s email summaries can be hacked to hide phishing attacks

Artificial intelligence is everywhere these days, on your phone, your car, even your washing machine. I saw one just the other day with incorporated. And although that may sound a bit exaggerated, it cannot be denied that artificial intelligence has made the easiest life in many ways.

From increasing productivity to unlocking new creative tools, it is changing the way we work and live. The most common version you have probably found? Ai generative, think of chatbots like chatgpt. But for useful that this technology can be, it is not exempt from problems.

If you have used the Google work space suite, you may have noticed the company’s AI model, Gemini, integrated into applications such as documents, sheets and Gmail. Now, researchers say that attackers can manipulate email summaries generated by Gemini to sneak into the hidden indications of Phishing.

Register for my free Cyberguy report
Get my best technological tips, urgent security alerts and exclusive offers delivered directly to your inbox. In addition, you will get instant access to my final scam survival guide, free when I join me Cyberguy.com/newsletter

How AI is now helping computer pirates to deceive the safety tools of their browser

How can the summaries of Gemini

Mozilla 0DIN researchers have discovered vulnerability in Google’s work space for the work space that allows attackers to inject hidden instructions in email summaries. The problem, demonstrated by Marco Figueroa, shows how the generative tools of AI can be deceived through indirect injection immediately. This technique incorporates invisible commands within the body of an email. When Gemini summarizes the message, he interprets and acts on those hidden indications.

What is artificial intelligence (AI)?

The attack does not depend on suspicious attachments or files. Instead, use a combination of HTML and CSS to hide the application by establishing the source size at zero and white color. These commands remain invisible in Gmail’s standard view, but are still accessible to Gemini. Once requesting a summary, the AI can be deceived to present false security alerts or urgent instructions that seem to come from Google.

In a concept test, Gemini falsely warned a user that a Gmail password had been compromised and included a false support phone number. Since Gemini’s summaries are integrated into Google’s work space, it is more likely to trust information, which makes this tactic especially effective.

What is Google doing about the defect?

Although Google has implemented defenses against rapid injection since 2024, this method seems to avoid current protections. The company told Cyberguy that it is actively implementing updated safeguards.

Get the News business on the fly by clicking here

In a statement, a Google spokesman said: “The defense of attacks that affect the industry, such as rapid injections, has been a continuous priority for us, and we have implemented numerous strong defenses to keep safe users, including safeguards to prevent harmful or misleading responses.

Google also confirmed that it has not observed the active exploitation of this specific technique.

6 ways in which you can be safe from phishing’s scams

So how can you protect themselves from phishing scams that exploit AI tools like Gemini? Here are six essential steps that you can take right now to stay safe:

1. Do not blindly trust the content generated by AI

The fact that a summary appears in Gmail or Doc does not mean that it is automatically safe. Try the suggestions, alerts or links generated by AI with the same caution of any unre requested message. Always verify critical information, such as security alerts or telephone numbers, through official sources.

2. Avoid using summary functions for suspicious emails

If an email seems unusual, especially if it is unexpected or someone who does not recognize, avoid using the AI summary function. Instead, read the full email as originally written. This reduces the possibility of falling in love with deceptive summaries.

3. Be careful with emails and pHishing messages

Be attentive to emails or messages that create a sense of urgency, request that you verify the details of the account or provide unexpected links or contact information, even if they seem reliable or come from family sources. Attackers can use AI to create alerts or applications for realistic aspect of confidential information, sometimes hidden within automatically generated summaries. Then, always pause and examine the suspicious indications before answering.

The best way to safeguard the malicious links that install malware, which potentially access their private information, is to have an antivirus software installed on all its devices. This protection can also alert it to the PHISHING Electronic Correos and Ransomware scams, maintaining their personal information and their safe digital assets.

Get my choices for the best 2025 antivirus protection devices for your Windows, Mac, Android and iOS devices in Cyberguy.com/Lockupyoutech

4. Keep your applications and updated extensions

Make sure Google Workspace and your browser are always running the latest version. Google regularly publishes security updates that help prevent new types of attacks. In addition, avoid using unofficial extensions that have access to their Gmail or documents.

5. Invest in a data removal service

The scams driven by AI as the Gemini summary attack does not occur in a vacuum. They often begin with stolen personal information. These data can come from past violations, public records or details that without knowing it has shared online. A data removal service can help continuously and requesting the elimination of your information from Data Broker sites. While no service can erase everything, reduce your fingerprint makes it more difficult for attackers to customize phishing attempts or link it to known non -compliance data. Think about it as another layer of protection in a world where AI makes specific scams even easier.

6. Disable AI summaries for now if you are worried

If you are concerned about a phishing attempt generated by AI, consider disabled the summaries of Gemini in Gmail until Google sets stronger protections. You can still read the complete emails in the traditional way, which can reduce your risk of being deceived by manipulated summaries.

How to disable Gemini functions on the desktop

• Open Gmail On the desk.
• Click on the Settings Gear icon in the upper right corner.
• Click See all configurations.
• Move to “Intelligent characteristics of the Google work space” and click Manage configurations of intelligent work space characteristics.
• Deactivate The lever for Intelligent characteristics in Google’s work space.
• Then, click Save.
• Note: This will Turn off the summaries of Gemini as well as other intelligent characteristics.

How to disable Gemini functions in Mobile

On iPhone:

If you use the gemini mobile application specifically:

• Open the Gemini application.
• Touch you Profile photo.
• Tap Gemini Application Activity.
• At the top, touch Turn off.

In Android:

SEthings can vary according to the manufacturer of your Android phone

• Open the Gmail application In your Android.
• Touch the Menu icon (three horizontal lines) in the upper left corner.
• Move down and touch Settings.
• Select the relevant email account.
• Move down and touch Intelligent characteristics of the Google work space and Unmarked the box To turn them off.

Key warnings to know:

• Disabling Intelligent characteristics You can eliminate other convenient features, such as predictive text and automatic appointment detection.
• He Gemini icon or summary buttons may still appeareven after disableing these characteristics. Some users report having to hide them physically through browser tools.

There is no unique “off -cut switch” to completely eliminate all references from Gemini everywhere, but these steps significantly reduce the presence and risk of the characteristic.

Click here to get the News application

Kurt key takeway

This vulnerability highlights how Phishing’s tactics evolve together with AI. Instead of relying on visible red flags, such as Mally written URL or suspicious attachments, attackers are now aimed at trust systems that help users to filter and interpret messages. As AI integrates more deeply into productivity tools, rapid injection could arise as a subtle but powerful vector for social engineering, hiding the malicious intention in the same tools designed to simplify communication.

How comfortable is letting you summarize or leak your emails, and where do you draw the line? Get us knowing in Cyberguy.com/contact

Register for my free Cyberguy report
Get my best technological tips, urgent security alerts and exclusive offers delivered directly to your inbox. In addition, you will get instant access to my final scam survival guide, free when I join me Cyberguy.com/newsletter

Copyright 2025 Cyberguy.com. All rights reserved.