Vulnerabilities
Chrome Sandbox Escape: AI-Assisted Chain of Three Zero-Days Impacts Windows Users
Xbow researchers used an AI model to chain three Chrome flaws, escaping the Windows sandbox in six days and earning a $250,000 bounty from Google.
Xbow researchers used an AI model to chain three Chrome flaws, escaping the Windows sandbox in six days and earning a $250,000 bounty from Google.
LabRedesCefetRJ released version 3.8.6 to fix a critical SQL injection flaw that allowed attackers to steal or delete data without logging in.
Malformed IPv4 packets targeting Secure Vector Routing interfaces allow attackers to crash Juniper SRX firewalls via a critical buffer overflow vulnerability.
An authorized attacker can exploit improper link resolution in .NET to elevate privileges locally on Linux and macOS systems running affected runtime versions.
A stale cache bug in Linux vhost allows local users to gain root privileges or lets virtual machines break out to the host system.
Microsoft patches a high-severity vulnerability in .NET WPF that allows attackers to run arbitrary code if users engage with crafted content.
Microsoft patches high-severity heap corruption flaw in .NET WPF runtime that enables remote attackers to execute code via crafted pixel-shader bytecode.
Unauthenticated requests can exhaust memory via synthetic culture names in ASP.NET Core, leading to denial of service across multiple platforms.
A high-severity deserialization flaw in .NET WPF lets attackers run code on Windows systems if a user interacts with a crafted link.
Threat actor UTA0576 chained three Ivanti zero-days to bypass authentication and execute commands as root on fully patched remote access gateways.
A critical flaw in the Samsung platform allows OS command injection, rated 9.8. No patch status is provided in the source material.
Default server configurations often serve file indexes when index files are missing, handing attackers a map of your internal structure without authentication.
A single injected property can overwrite default settings in a browser, turning safe default behaviors into security failures without any user interaction.
A critical code injection flaw in Adobe ColdFusion allows remote attackers to run arbitrary code without needing victim interaction.
SQL injection targets the database interpreter, while command injection targets the operating system shell, creating distinct attack surfaces that require separate mitigation strategies.
API flaws often bypass traditional web defenses by exposing data logic directly, allowing attackers to manipulate business rules rather than just stealing credentials.
Attackers with physical access or valid credentials can bypass UEFI Secure Boot on Cisco UCS servers using shell memory commands.
Virtual patching stops attacks at the network edge by filtering malicious traffic, allowing you to protect systems immediately while developers prepare permanent code fixes.
A critical buffer overflow in Network Policy Server lets attackers run code remotely on affected Windows versions without prior authentication.
An unauthorized attacker can execute code over a network via a critical use-after-free vulnerability in the Microsoft QUIC protocol on multiple Windows 11 versions.
The unsafe-inline directive nullifies Content Security Policy protections by permitting any script execution, turning a security header into a decorative label that attackers bypass with ease.
A critical flaw in IBM’s MCP Context Forge lets attackers create admin tokens without credentials. The issue persists until version 1.0.2 is deployed.
Attackers could hijack over 1,500 npm packages by reclaiming abandoned GitHub repository names, bypassing credential checks to push updates.
Most self-hosted GitLab breaches stem from misconfiguration and exposed services, not zero-day exploits in the core codebase.