Description

Security researchers have discovered a new attack, named TunnelVision, that compromises the security of nearly all virtual private network (VPN) applications. This attack forces VPNs to send and receive traffic outside of the encrypted tunnel designed to protect it, effectively negating the purpose of VPNs to cloak user IP addresses and encrypt internet traffic. The attack, which may have been possible since 2002 and could have already been used in the wild, affects all VPN applications except those running on Linux or Android. TunnelVision works by manipulating the DHCP server that assigns IP addresses to devices connecting to the local network. By using a setting known as option 121, the attacker can override default routing rules and divert VPN traffic through the DHCP server itself, rather than through the encrypted tunnel. As a result, the attacker gains control over the victim's traffic, allowing them to read, drop, or modify the leaked data while the victim remains connected to both the VPN and the internet. The attack can be carried out by someone with administrative control over the network or by setting up a rogue DHCP server. While Android is immune to the attack as it doesn't implement option 121, other operating systems have no complete fixes. Even on Linux, where there is a partial mitigation setting, TunnelVision can still be used to exploit a side channel for de-anonymizing destination traffic and performing targeted denial-of-service attacks. Network firewalls can be configured to mitigate the attack, but this is problematic for VPN users on untrusted networks, as they have no control over the firewall settings. The most effective fixes against TunnelVision include running the VPN inside a virtual machine with a network adapter not in bridged mode, or connecting the VPN to the internet through a cellular device's Wi-Fi network. Researchers from Leviathan Security, Lizzie Moratti and Dani Cronce, detailed the attack in a research paper.