The Wire
Curated cybersecurity and tech news — AI-summarized, source attributed.
StreetComplete iOS Beta Launches, Seeks Contributors
StreetComplete, an open-source app for improving OpenStreetMap data, has entered public beta for iOS. The project uses Kotlin Multiplatform to share code betwee
Threat actors weaponize custom GPTs to deliver RATs
Cybercriminals are creating malicious custom GPTs on ChatGPT to trick users into downloading remote access trojans (RATs). These custom GPTs appear legitimate,
TLA+ Limitations Exposed Beyond Hype
TLA+ is a powerful tool for designing complex concurrent systems and verifying safety properties, but it has significant limitations. While it excels at checkin
Trump, Tech Giants Agree on Voluntary AI Safety Pledge
President Trump and major tech companies have reached a voluntary agreement to enhance AI safety. The accord aims to ensure responsible development and deployme
Turbopuffer Abandons Vector-First Architecture for Better Performance
Turbopuffer, a serverless vector database, is overhauling its storage architecture to improve search performance across text, regex, and vector queries. The com
Weekly Cyber Threats: AI, Secrets, and Evasion Tactics
This week's cyber threats highlight how attackers exploit ordinary system functions like model checks and caching to launch attacks. They often reuse old flaws
Western AI Firms Quietly Adopt Chinese Optimizations
Western AI companies are integrating Chinese-developed KV cache optimizations into their models without public acknowledgment. DeepSeek's breakthroughs, which r
Zero Trust's Weak Spot: Day-One Identity Verification Flaws
Zero Trust architectures focus on securing existing identities but often overlook vulnerabilities during onboarding. Human error and weak initial identity check
Zimbra flaw exploited to steal emails, install web shells
Hackers are exploiting a critical Zimbra Collaboration Suite vulnerability (CVE-2026-73570) to steal emails and authentication credentials. The flaw allows remo
101 npm packages secretly add devs' WhatsApp to groups
Researchers found 101 malicious npm packages exploiting the Baileys WhatsApp project to add developers' accounts to groups without consent. The packages, collec
543,000 Valid Credentials Still Exposed on GitHub Despite Safeguards
Truffle Security's analysis of 224 million GitHub repositories found over 543,000 valid credentials exposed, with some remaining accessible for up to 784 days.
6 Dangerous Browser-Based Attack Techniques in 2026
Browsers are prime targets for attacks, with most breaches starting and ending in browser sessions. Six key threats in 2026 include phishing for credentials and