TechCapsules

A new cyberattack campaign, dubbed ClickFix, has been targeting Windows users through fake CAPTCHA pages since April 2026. The campaign combines three techniques - EtherHiding, a blockchain-based payload method; ClickFix, a social engineering trick; and GULoader, a memory-based malware downloader - to evade standard security defenses. The attack begins on a compromised European small-business website and ends with an attempt to load GULoader onto a victim's machine. The campaign targets users who arrive at a legitimate-looking website [...]

Microsoft claims its Defender catches nearly all malicious email on its own, with integrated third-party tools improving detections by less than 1%. The company's quarterly benchmarking data shows Defender misses the fewest malicious emails compared to competitors and removes nearly 100% of dangerous emails that reach the inbox. However, security experts urge caution, stating that percentages obscure the true quantity and severity of emails that get through. Microsoft's data ranks itself against other security vendors, including Mimecast [...]

A security researcher discovered a vulnerability in FIFA's internal system, allowing her to access and control the TV stream of every World Cup game. By registering as a player agent on FIFA's official platform, she exploited a flaw in the back-end API, gaining access to internal systems, including the one controlling TV broadcasts. The researcher reported the issue, and FIFA fixed it within hours without acknowledging her report.

EigenQ and TD SYNNEX have partnered to secure AMD EPYC server environments against post-quantum threats. The collaboration aims to advance post-quantum cryptographic readiness across high-assurance environments, including public sector, defense, and critical infrastructure. The joint framework combines EigenQ's quantum-safe trust infrastructure with AMD EPYC's hardware-level security architecture, providing scalable migration pathways to protect data. The integration addresses vulnerability mitigation and ensures alignment with federal CNSA 2.0 timelines and mandates, targeting newly procured and existing enterprise server ecosystems [...]

The vulnerability crisis has escalated with the emergence of machine-speed AI threats, where models like Anthropic's Claude Mythos Preview can autonomously exploit vulnerabilities with an 83.1% success rate. Human-speed patching cannot keep up, and CIOs must implement a strict framework to secure the logic horizon, the point where AI models turn natural language into business-critical actions. The three-layer control pivot includes execution control, identity and dynamic authorization, and data governance to lock down the logic horizon and [...]

NTT DATA's Insurtech Global Outlook 2026 report highlights the insurance sector's approaching inflection point due to growing cyber and climate-related losses. The report estimates uninsured cyber losses to rise from $171 billion in 2023 to over $700 billion by 2030, with climate-related uninsured losses at $180 billion. Insurers struggle to scale artificial intelligence safely and consistently, despite 66% of employees using AI tools. The report identifies key shifts shaping the insurance industry and provides guidance on building [...]

Google has taken down a major AI scam ring called "Outsider Enterprise," accusing the group of using its Gemini AI and other services to fuel a massive phishing operation. The operation, which targeted over 9,000 phony websites and over a million scam URLs, resulted in almost 55,000 reported spam texts and 2.5 million malicious messages sent to Android devices. Google estimates hundreds of thousands of people have been impacted, with financial losses in the millions of dollars.

A critical vulnerability chain has been discovered in LiteLLM, a widely deployed open-source AI gateway. The chain, rated CVSS 9.9, allows low-privilege users to climb to full admin and run code on the server, exposing sensitive information such as provider keys, credentials, and prompts. The vulnerability affects LiteLLM v1.83.13 and earlier, and a fix is available in v1.83.14-stable, released on May 2. Users are advised to upgrade to the latest version to close the three-CVE chain.