Stories cluster related articles into a single narrative, linked to the underlying CVEs and affected products. 8,564 stories synthesized.
New research introduces 'Perturbation Probing,' a method to diagnose LLM safety vulnerabilities, finding that AI safety mechanisms reside in a thin neural layer susceptible to manipulation.
A critical balance-handling flaw in the Cosmos EVM module was exploited to drain funds from six blockchains, occurring after Cosmos Labs was aware of the widespread vulnerability.
The Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) has confirmed a cyberattack impacted a standalone system containing information on investigation targets, with the ransomware group Qilin claiming responsibility.
Key findings • Four medium-severity vulnerabilities disclosed together in GIMP image parsing plugins. • Flaws include heap out-of-bounds reads and stack out-of-bounds accesses in file-psd, fi…
Key findings • 16 CVEs disclosed between Aug 25-28, 2026, affecting GIMP, Nokogiri, Dia, file-roller, libsoup, and gdk-pixbuf. • Multiple GIMP image parsing plugins vulnerable to DoS via heap…
Key findings • Microsoft Edge: One high and six medium severity vulnerabilities disclosed on August 28, 2026. • Vulnerabilities include type confusion, LLM prompting issues, origin validation…
Kidney dialysis provider DaVita will pay $15 million to settle class action lawsuits stemming from a 2025 ransomware and data theft incident attributed to the Interlock gang, impacting nearly 2.7 million individuals.
An international wave of law enforcement actions has targeted major cybercrime syndicates and state-sponsored hacking groups, leading to arrests, asset seizures, and disruption of critical infrastructure espionage.
Researchers have identified 19 malicious browser extensions for Chrome and Edge containing code designed to steal cryptocurrency wallet secrets and drain funds.
The TITAN ransomware group claims to employ an AI system capable of analyzing 700GB of stolen data per hour, significantly accelerating its extortion tactics.
A former Defense Intelligence Agency IT specialist has pleaded guilty to leaking classified information to a foreign government after being caught in an FBI sting operation.
The looming threat of quantum computers breaking current encryption necessitates proactive cryptographic inventory and migration to quantum-resistant algorithms, with regulatory deadlines fast approaching.
A widespread phishing operation is using SVG files disguised as voicemails to evade email security, impacting thousands of organizations.
Key findings • Eight vulnerabilities disclosed for Budibase on August 28, 2026, patched in v3.41.3. • Critical RCE vulnerability (CVE-2026-82244) allows arbitrary code execution via malicious…
Key findings • 17 Dovecot vulnerabilities disclosed on August 28, 2026, ranging from Low to Critical severity. • Critical flaw CVE-2026-42007 involves use-after-free in mail editing via Sieve…
Security researcher Olivier Laflamme has disclosed two independent vulnerabilities, CVE-2026-76639 and CVE-2026-76640, that allow root remote code execution on Unitree G1 EDU humanoid robots, with one flaw being exploitable over Bluetooth Low Energy.
A sophisticated phishing campaign is targeting academic researchers with a fake student resume designed to deliver the SNOWLIGHT and VShell remote-access trojans.
Rubrik CEO Bipul Sinha highlights a growing enterprise demand for integrated platforms that manage AI agent identities, provide visibility into their actions, and enable robust recovery capabilities.
CISA's latest review reveals that common, long-standing vulnerability classes like injection flaws and improper input validation remain the most exploited, highlighting a critical gap in software development practices.
ServiceNow has released patches for four critical vulnerabilities in its AI Platform, three of which carry a CVSS score of 10.0 and could allow unauthenticated attackers to execute arbitrary code and perform SQL injection attacks.
Suspected North Korean remote workers are increasingly infiltrating legitimate companies by posing as sales, marketing, and even medical professionals, presenting a novel challenge for cybersecurity defenses.
Mathematicians and AI researchers convened to assess AI's current and future role in mathematical discovery, concluding that while AI excels at problem-solving, true theoretical novelty remains a human domain, for now.
Two undocumented factory implants, SPEAKINGSTONE and DARKLANTERN, discovered in ZBT router firmware allow unauthenticated attackers to gain root access and execute commands.
Leaked documents from Bauman Moscow State Technical University reveal a GRU cyber training program that appears to feed personnel into notorious APT28 and Sandworm units.