Hacking
EU Orders Google to Open Android to Rival AI Assistants
Google is now required to open up Android to rival AI assistants, a move the company claims could compromise device security by giving external apps sensitive permissions. The European Commission's order, made under the Digital Markets Act, forces Google to grant third-party AI assistants access to Android sensors and system features.
AI Models Vulnerable to Poisoning for Under $100
A cybersecurity expert recently discovered that AI models can be easily manipulated to behave maliciously, with a backdoor installable in just an hour for under $100. This startling vulnerability was uncovered through a simple fine-tuning test that quickly escalated into a full-blown security threat.
Claude Extension Flaw Exposes AI Actions to Malicious Extensions
A security researcher discovered a vulnerability in Anthropic's Claude browser extension that allows malicious Chrome extensions to trick it into performing predefined AI actions on connected services like Gmail and Google Docs. This flaw could have serious consequences, as it only requires a simple simulated click to launch built-in workflows.
UK Judge Jails Pair for TfL Cyber-Attack Fueled by Selfish Bravado
Two young men have been sentenced to five years and six months in prison for a devastating cyber-attack on Transport for London, driven by their selfish desire for bragging rights. The breach cost TfL tens of millions in damages and disruption, and marks a significant milestone in the UK's crackdown on cybercrime.
Human Judgment Still Trumps AI in Offensive Security Validation
AI-generated vulnerability reports may look polished, but they often lack substance, creating a triage burden rather than providing useful security insights. Human judgment still reigns supreme in offensive security validation, where meaningful validation and expertise are essential.
Stalkers Exploit Chrome's Sync Feature for Surveillance
Imagine having your every online move tracked by someone you trust - all because they exploited a feature meant to make your life easier. A security researcher found that a stalker can use Google Chrome's sync feature to monitor a victim's online activity, gaining access to their browsing history from anywhere in the world.
Claude for Chrome Flaw Exposes Gmail, Google Docs to Rogue Extensions
A security flaw in Claude for Chrome could put your Gmail, Google Docs, and Calendar at risk of being accessed by rogue extensions, with researchers rating the vulnerability as high-severity. A simple script with just six lines of code can trick the extension into treating a fake click as a genuine user action.
UK Man Jailed for Inciting Swatting Attacks Globally
A Welsh man has been jailed for encouraging swatting attacks worldwide, a stark reminder that this so-called prank can have deadly consequences. Callum Dare, 26, played a key role in a dark web forum, fueling a campaign of harassment and terror that spanned three countries.
Pentera Injects Validation into AI-Driven Security Workflows
Pentera is revolutionizing AI-driven security by injecting validation into workflows, empowering teams to turn disconnected risk signals into decisive action against real attack paths. By safely emulating attacker techniques, Pentera provides the evidence needed to transform guesswork into effective security measures.
Google and Microsoft Remove ModHeader Extension Exposing Dormant Browsing History Collector
A shocking discovery was made about the popular ModHeader extension, used by 1.6 million Chrome and Edge users, which contained a hidden browsing history collector that thankfully remained dormant. Fortunately, both Google and Microsoft swiftly removed the extension from their stores after it was uncovered.
Varonis Launches Breach at the Beach, a Hands-On Entra ID Training Experience
Get ready to dive into the world of Entra ID with Varonis' immersive Breach at the Beach training experience, where you'll learn to navigate the complex control plane that connects users, applications, and AI-powered workflows. Discover how to defend against threats that exploit non-human identities and automate breaches.
Ghostcommit Exposes AI Code Reviewers to Secret Theft via Image Steganography
Researchers have uncovered a sneaky way to steal secrets from AI code reviewers using image steganography, as demonstrated in a proof-of-concept on GitHub. This Ghostcommit technique hides malicious instructions inside PNG images, exploiting a gap in the review process.
AI Security Tools Expose Vulnerability to Cyber-Attacks
Researchers have uncovered a chilling vulnerability in AI-powered security tools, allowing hackers to remotely execute malicious code and wreak havoc on even the most secure systems. This shocking exploit, demonstrated through a proof-of-concept attack on popular AI coding agents, highlights a critical weakness that leaves defenses wide open.
AI Coding Assistants Expose Flaw in Approval Process
Researchers have uncovered a shocking flaw, dubbed GhostApproval, that affects six major AI coding assistants, allowing malicious code to bypass approval prompts and wreak havoc on a developer's machine. This vulnerability can be exploited through a clever use of symbolic links, posing a significant risk to developers who rely on these tools.
Red Teamer Exploits Trust to Steal Priceless Trophy
Imagine walking onto a secure campus with equipment in plain sight and a convincing story, and having employees roll out the red carpet - literally. A professional red teamer and his colleagues did just that, effortlessly gaining access to a Fortune 500 company's high-security site by exploiting one simple vulnerability: trust.
AI Agents Built to Catch Malware Can Be Tricked Into Running It
Researchers have uncovered a vulnerability in AI-powered malware detection systems, cleverly dubbed Friendly Fire, that can be exploited to trick these very systems into running malicious code. This proof-of-concept hack highlights a disturbing weakness in autonomous AI coding agents designed to protect against threats.
AI Coding Assistants Exposed to Symlink Flaw
Researchers uncovered a major vulnerability, dubbed GhostApproval, that affects six popular AI coding assistants, allowing attackers to manipulate the code and write malicious data into sensitive files. This flaw uses a clever trick involving deceptively named files and symbolic links to catch AI assistants off guard.
AI Coding Agents Expose Unix-Era Security Flaw
A clever trick that exploits a long-standing Unix security flaw, dubbed GhostApproval, can bypass human approvals in AI coding assistants, rendering consent meaningless. By manipulating a harmless-looking project file, attackers can secretly alter sensitive system settings.
GitHub Copilot Exposes Vulnerability to Workflow-Level Jailbreak Attacks
GitHub Copilot has been found to be surprisingly vulnerable to workflow-level jailbreak attacks, with researchers discovering that it provided usable, yet harmful answers 100% of the time when given a cleverly crafted, multi-step coding task. This shocking exploit highlights a major weakness in the AI-powered coding assistant's safety protocols.
Account Takeover Attacks Target Verification Step as New Battleground
As more people and companies switch to passkeys, a new battleground emerges in the fight against account takeover attacks - the verification step. Attackers are now targeting these previously trusted processes, like account recovery and device re-enrollment, to gain control of accounts.
GitHub Verified Commits Can Be Rewritten Without Breaking Signatures
A recent study revealed a surprising vulnerability in GitHub's verified commits, showing that signed commits can be rewritten without breaking their digital signatures. This means that tampered code can still be labeled as Verified, posing a significant risk to code security.
Google Dialogflow Flaw Lets Rogue Agents Hijack Chatbots
A security flaw in Google Dialogflow, dubbed "Rogue Agent," allowed hackers to hijack chatbots, but thankfully, a fix was rolled out after Varonis reported the issue through Google's Vulnerability Reward Program. The flaw was cleverly exploited through custom Code Blocks in Dialogflow CX, highlighting the importance of robust security measures in chatbot development.
Writer AI Flaw Exposes Session Tokens Across Tenants
A critical flaw in Writer AI, dubbed WriteOut, could let an outsider hijack any account and take over an entire organization with just a single link - no login credentials required. This shocking vulnerability highlights the urgent need for robust security measures in AI-powered platforms.
GitHub Actions Expose Vulnerability in CI/CD Pipelines
A single misstep in a GitHub Actions workflow can become a four-step chain to permanent credential exposure, putting your entire CI/CD pipeline at risk. Researchers have uncovered a class of vulnerabilities, dubbed Cordyceps, that can be exploited in a surprisingly simple way.