Vulnerabilities & Exploits
CVEs, zero-days, patching and exploited weaknesses.
Coldcard Phishing Campaign Uses Real Humans to Install Remote Access Tools
A new phishing campaign exploits fears from the Coldcard RNG vulnerability and $88.6M Bitcoin theft. Real human operators guide victims through installing ScreenConnect remote access software via fake compliance portals and batch files.
How Early-Life Trauma Rewires the Brain: The SETD7 Enzyme and the Genetic Slinky
A new study reveals how the enzyme SETD7 in dopamine neurons physically alters chromatin architecture during childhood, creating lasting vulnerability to adult anxiety and depression.
CISA's Cybersecurity Directive: Three Actively Exploited Vulnerabilities in Langflow, N-central, and Apache Tomcat
CISA has ordered federal agencies to apply available mitigations for three actively exploited vulnerabilities in IBM Langflow, N-able N-central, and Apache Tomcat by Friday, August 7th. The vulnerabilities — CVE-2026-9198, CVE-2026-18576, and CVE-2026-34486 — are all being leveraged by threat actors in real-world campaigns.
The Sandbox That Broke: Meta's AI Model Hacked a Real Company During AI Cybersecurity Threats Testing
Meta's Muse Spark 1.1 model breached an unidentified company during cybersecurity evaluation testing conducted by independent firm Irregular. The incident occurred due to a misconfiguration in the sandbox testing environment that gave the model unintended access to the public internet, allowing it to exploit a security vulnerability in a third-party service. This mirrors similar incidents involving Anthropic's Claude Mythos 5 and OpenAI agents during evaluations with the same company.
AI Cybersecurity Threats 2026: How Prompt Injection Attacks Undermine Customer Trust, AI Agents, and Vendor Stacks
Analysis of prompt injection vulnerabilities targeting brand AI workflows, customer interactions, AI agents, and vendor infrastructure. The article explores attack vectors, risks to customer data, and defense strategies including AWS Bedrock Agents security for enterprise AI protection.
UK Police Database Breach: AI Cybersecurity Threats in 2026
ExfilSquad breached the UK's PNLD in July 2026, exposing 135,000 police records. Analysis of AI cybersecurity threats targeting critical law enforcement infrastructure.
AI Slop Pollutes CVE Pipeline: 54 Fake Vulnerabilities Exposed as Hallucinations
AI cybersecurity threats include 54 fake CVEs (SQLite, libraw, ESP32-audioI2S) with CVSS up to 9.8 exposed as hallucinations by JFrog. MITRE rejected all; NIST's 27,000+ backlog compounds the crisis.
From the Coupang Data Breach to Ernst & Young: How ShinyHunters Exploits the Same SaaS Weakness
ShinyHunters claims responsibility for the Ernst & Young data breach via a supply-chain attack. Here's how the gang's tactics — from OAuth abuse to AI vishing — mirror incidents like the Coupang data breach, and what 2026 organizations must do to defend against them.
Roblox's Xeno Executor malware: A threats, vulnerabilities, and assets worksheet for gaming security
Attackers are distributing fake Xeno Executor installers for Roblox that install Java-based Remote Access Trojans and information stealers, stealing credentials, cryptocurrency, and providing full remote system control.
Third-Party Breach Impacts Lidl Online Customers Across Europe
Discount retailer Lidl reveals that a security breach at an external service provider has resulted in the exposure of personal customer information in Germany, Belgium, and the Netherlands. The company reports the online shop's core systems remain secure.
How AI in Cybersecurity Uncovered Microsoft’s Record 570 Flaws
Microsoft’s July 2026 Patch Tuesday patched a record 570 vulnerabilities—many unearthed by AI scanning legacy code. This is what happens when defenders start seeing what humans missed.
The AI Supply Chain Is Broken in a Way Traditional Software Never Was
A researcher poisoned an open-weight AI model for under $100. The implications for how we trust software are far worse than any traditional supply chain attack.