- Computer pirates are using invisible unicode to deceive Android to open dangerous notification links
- The link looks normal, but Android secretly opens something more without prior notice or consent
- Even trust applications such as WhatsApp and Instagram are vulnerable to this hidden notification exploitation
A safety failure in the Android notification system could allow malicious actors to deceive users to open unwanted links or activate hidden application actions, experts warned.
IO-No investigation states that the fault lies in how Android analyzes certain unicode characters within notifications.
This creates a mismatch between what users see and what the system processes when the “open link” suggestion appears.
What you see is not always what you get
The problem comes from the use of invisible or special Unicode characters within the URLs.
When included in a message, these characters can make Android interpret the visible text and the real action link differently.
For example, a notification can visibly show “Amazon.com”, but the underlying code actually opens “zon.com”, with a character of zero wide space inserted.
The notification is shown as “Ama[]Zon.com, “including the hidden character. However, the suggestion engine interprets that the hidden nature as a separator, resulting in the launch of a completely different site.
In some cases, attackers can redirect users not only to websites but also deep links that interact directly with applications.
The report showed how a seemingly harmless shortened URL led to a WhatsApp call.
To make the attacks less detectable, malicious actors can use URL shorters and embed the links in a reliable -looking text.
The fault becomes particularly dangerous when combined with applications or “deep links” that can silently activate behaviors, such as starting messages, calls or opening views of internal applications without user intention.
The tests on devices that include the Google Pixel 9 Pro XL, Samsung Galaxy S25 and the oldest Android versions revealed that this bad behavior affects the main applications such as WhatsApp, Telegram, Instagram, Discord and Slack.
Personalized applications were also used to avoid characters filtering and validate the attack in multiple scenarios.
Given the nature of this defect, many standard defenses can fall short. Even the best antivirus solutions can lose these exploits, since they often do not involve traditional malware discharges.
On the other hand, the attackers manipulate the behavior of the user interface and exploit the application link settings. Therefore, there is the need for final point protection tools, which offer broader detection based on behavioral anomalies.
For users at risk of theft of credentials or application abuse, relying on identity theft protection services becomes critical to monitor unauthorized activity and ensure exposed personal data.
Until a formal solution is implemented, Android uses must remain cautious with notifications and links, especially those of un familiar or short -comparators of URL.
