Malware

Twitter

Facebook

Copy Link

Top News

Amazon Q security flaw let malicious Git repos execute code and steal AWS credentials via MCP

A high-severity security flaw in Amazon Q Developer allowed attackers to execute arbitrary code and steal cloud credentials through booby-trapped Git repositories. The vulnerability, tracked as CVE-2026-12957 with a CVSS score of 8.5, exploited how the AI coding assistant handled Model Context Protocol configurations. Wiz Research discovered the bug and Amazon patched it, but the incident highlights broader supply chain risks across AI developer tools.

Technology3 sources

Amazon Q security flaw let malicious Git repos execute code and steal AWS credentials via MCP

Technology3 sources

North Korean Hackers Deploy Gaslight Malware to Deceive AI Security Tools with Fake Errors

A newly discovered macOS malware called Gaslight uses prompt injection to confuse AI-assisted analysis platforms. Attributed to North Korea-aligned threat actors, the Rust-based implant embeds 38 fabricated system messages designed to make AI triage agents abort or refuse analysis. The malware also functions as an information stealer with Telegram-based command-and-control capabilities.

Technology2 sources

North Korean Hackers Deploy Gaslight Malware to Deceive AI Security Tools with Fake Errors

Technology2 sources

Fake AI agent skill bypassed every security scanner and reportedly reached 26,000 agents

Security firm AIR created a fake AI agent skill that passed every major security scanner and reportedly reached 26,000 agents, including corporate accounts. The experiment exposed critical flaws in how AI agent skills are vetted, showing that scanners only check submitted packages while attackers can modify external links after approval. The demonstration highlights urgent security gaps in the rapidly growing AI agent ecosystem.

Technology2 sources

Fake AI agent skill bypassed every security scanner and reportedly reached 26,000 agents

Technology2 sources

Five Eyes Alliance Warns AI Cyber Risks Could Trigger Crises Within Months, Not Years

Intelligence leaders from the US, UK, Canada, Australia, and New Zealand issued an urgent joint warning that frontier AI models will fundamentally transform offensive cyber capabilities within months. The statement calls for immediate action on cybersecurity fundamentals as AI-powered cyber threats accelerate the speed and sophistication of attacks, potentially causing major operational and financial crises.

Policy16 sources

Five Eyes Alliance Warns AI Cyber Risks Could Trigger Crises Within Months, Not Years

Policy16 sources

Cybercrime accounts for 30% of all crime in Asia-Pacific as scam centers generate $40 billion

Interpol's latest assessment reveals a dramatic shift in the regional crime landscape, with cybercrime now representing nearly a third of all recorded offenses across Asia and South Pacific. The surge is driven by industrial-scale scam centers generating close to $40 billion annually, sophisticated AI-driven fraud, and organized criminal networks exploiting rapid digitalization faster than law enforcement can respond.

Policy3 sources

Cybercrime accounts for 30% of all crime in Asia-Pacific as scam centers generate $40 billion

Policy3 sources

Malicious JetBrains plugins steal AI API keys from 70,000 developer installations

Cybersecurity researchers at Aikido Security uncovered at least 15 malicious plugins on JetBrains Marketplace designed to steal AI API keys from developers. The coordinated malware campaign includes AI coding assistants like DeepSeek AI Assist and CodeGPT AI Assistant, with nearly 70,000 combined installations. The plugins function as advertised but secretly exfiltrate API keys to attacker-controlled servers, potentially reselling stolen credentials to paying users.

Technology2 sources

Malicious JetBrains plugins steal AI API keys from 70,000 developer installations

Technology2 sources

Chinese-linked hackers ran year-long cyberespionage campaign against U.S. and Canadian research facilities

Google's Threat Intelligence Group uncovered a sophisticated cyberespionage campaign that silently infiltrated U.S. and Canadian research institutions for over two years. Chinese-linked hackers stole sensitive data on defense intelligence, artificial intelligence, and medical research by exploiting vulnerabilities in widely-used REDCap servers. The campaign targeted organizations with billions in combined research budgets.

Policy3 sources

Chinese-linked hackers ran year-long cyberespionage campaign against U.S. and Canadian research facilities

Policy3 sources

Agentjacking attack tricks AI coding agents into executing malicious code via fake bug reports

Cybersecurity researchers at Tenet Security have uncovered Agentjacking, a novel attack method that hijacks AI coding agents through fake Sentry error reports. The attack exploits Model Context Protocol to execute malicious code on developer machines with an 85% success rate, bypassing traditional security measures like EDR and firewalls. At least 2,388 organizations face exposure through vulnerable Sentry DSNs.

Technology2 sources

Agentjacking attack tricks AI coding agents into executing malicious code via fake bug reports

Technology2 sources

Developers Ship AI-Generated Code Despite Knowing It's Riddled With Security Vulnerabilities

A Checkmarx survey of 2,350 developers and security leaders reveals 70% believe AI-generated code contains more security vulnerabilities than human-written code. Yet pressure to deploy quickly means 30% knowingly ship vulnerable code into production. With AI now accounting for nearly half of all production code, organizations face a critical gap between development speed and security oversight.

Technology3 sources

Developers Ship AI-Generated Code Despite Knowing It's Riddled With Security Vulnerabilities

Technology3 sources

Hades malware tricks AI security scanners with fake nuclear weapon prompts to hide payloads

A sophisticated supply chain attack called Hades targets Python developers by deceiving AI-powered security tools with adversarial prompt injection. The malware embeds fake instructions about creating biological and nuclear weapons in code comments, triggering safety failsafes that cause scanners to skip the actual malicious payload. Security researchers have identified 37 Python and 106 JavaScript packages compromised in this campaign, which steals credentials from AWS, GitHub, Kubernetes, and AI developer tools.

Technology3 sources

Hades malware tricks AI security scanners with fake nuclear weapon prompts to hide payloads

Technology3 sources

Self-replicating Miasma worm compromises 73 Microsoft GitHub repositories in supply chain attack

Microsoft removed 73 GitHub repositories after they were infected with the Miasma worm, a self-replicating credential-stealing malware that activates when developers open projects in AI coding tools. The attack exploited the same compromised credentials from a May breach, raising questions about incomplete credential rotation and highlighting vulnerabilities in the modern software supply chain.

Technology9 sources

Self-replicating Miasma worm compromises 73 Microsoft GitHub repositories in supply chain attack

Technology9 sources

Google Gemini security flaw let hackers hijack Android phones via WhatsApp notifications

SafeBreach Labs uncovered a notification-based prompt injection vulnerability in Google Gemini on Android that allowed attackers to send hidden malicious commands through WhatsApp, Slack, and other messaging apps. The security flaw enabled silent control of smart home devices, memory poisoning, and surveillance without user interaction. Google has since deployed a server-side patch to block the exploit.

Technology2 sources

Google Gemini security flaw let hackers hijack Android phones via WhatsApp notifications

Technology2 sources

Scientists build AI computer worm that learns and adapts as it spreads across networks

Researchers at the University of Toronto have created a prototype AI computer worm that spreads autonomously across networks, learning and adapting its attacks with each infection. Built using freely available open-weight AI models, the worm infected half a test network in five days, exploiting known vulnerabilities across Windows, Linux, and IoT devices. Experts call it a wake-up call for cybersecurity preparedness.

Technology10 sources

Scientists build AI computer worm that learns and adapts as it spreads across networks

Technology10 sources

Popular OpenAI Codex tool with 29,000 weekly downloads exposed in npm supply chain attack

A malicious npm package posing as a remote UI for OpenAI Codex quietly exfiltrated developer authentication tokens for a month, attracting 29,000 weekly downloads. The same credential-theft chain ran through two Android apps with over 60,000 combined downloads. The attack highlights how threat actors can hide malicious code in npm builds while keeping GitHub repositories clean, exposing critical vulnerabilities in AI developer tooling.

Technology4 sources

Popular OpenAI Codex tool with 29,000 weekly downloads exposed in npm supply chain attack

Technology4 sources

Sophos MDR achieves 89-second threat response with AI-powered agentic SOC defending 40,000 customers

Sophos has revealed production results from a full year of agentic operation inside its Managed Detection and Response platform, now defending 40,000 customers worldwide with 39% year-over-year growth. The cybersecurity firm's AI-powered Security Operations Center neutralizes threats in just 89 seconds while autonomously closing 52% of cases without human intervention, setting a new benchmark for managed security operations at scale.

Technology2 sources

Sophos MDR achieves 89-second threat response with AI-powered agentic SOC defending 40,000 customers

Technology2 sources

Prev12 3 4…7 Next