ACTIVE THREATS: CISA: Hackers now exploit max severity GitLab flaw in attacks • How a hole in Lenovos login system let hackers walk into 5,000 Dropbox accounts • The US military just turned off ad tracking on its phones. Maybe you should too • Hackers exploit Tencent app flaw to deploy GrayRabbit malware • CRPx0 ransomware: what you need to know ACTIVE THREATS: CISA: Hackers now exploit max severity GitLab flaw in attacks • How a hole in Lenovos login system let hackers walk into 5,000 Dropbox accounts • The US military just turned off ad tracking on its phones. Maybe you should too • Hackers exploit Tencent app flaw to deploy GrayRabbit malware • CRPx0 ransomware: what you need to know
The full HackWire archive — 3,892 stories, newest first.
KARR anti-theft systems use a shared Bluetooth key across all devices, allowing attackers within 30 meters to unlock doors or disable engines (CVE-2026-18411, CVSS 8.1). No physical access to the vehicle required.
Thermo Fisher genetic analyzers lack integrity checks on DNA output files, allowing results to be silently falsified via file-system access (CVE-2026-17583, CVSS 8.4). Multiple product lines used in clinical and forensic labs worldwide have no available patches, including end-of-life products.
Organizations overlook AI's core security risk: data exfiltration through multi-prompt conversations that bypass CASB/DLP tools designed for SaaS. The exposure happens in dialogue, not access—sensitive info distributed across exchanges that trigger no pattern matches.
Greatness phishing kit now includes device code phishing, which steals authenticated session tokens after MFA completes. This nation-state technique, now commercially available to criminals, renders MFA-only security insufficient and makes account takeover possible without password theft.
Russ Kirby became a CISO by accident and succeeded. His cognitive fit for toggling between threat models and details, plus rare self-awareness, explains his resilience versus typical burnout.
Exploits happen in hours, making patching too slow. Oligo's $60M bet signals the industry's pivot to runtime security—the only way to know which vulnerabilities matter before attackers strike.
AI agents with broad system access lack human safeguards and follow instructions blindly—even corrupted ones from prompt injection. Varonis's Agent IBAC addresses this by verifying agents act as intended, not just according to role.
ChainDrop, an npm worm from a compromised Keyv maintainer account, infected 1,300+ packages (2B monthly downloads) and spread via preinstall hooks and credential theft. It defeated provenance attestation as a security defense.
Google's ADK automation workflows were compromised via prompt injection through GitHub issues, revealing that AI agents reading untrusted input while holding credentials are fundamentally exploitable.
Researchers showed email AI assistants can be exploited via prompt injection to steal data, impersonate executives, and enable fraud. Enterprises deploying these systems face critical security risks.
A Firebase misconfiguration in tl;dv exposed users' meeting recordings to any authenticated account holder, potentially allowing unauthorized access to confidential government and enterprise calls. The flaw exemplifies a recurring pattern of Firebase security oversights in production deployments.
AI eliminates the technical barrier to attacks. Motivated but unskilled actors—insiders, hacktivists, opportunists—can now generate working exploits via LLMs, reshaping who poses an enterprise threat.
Attackers use fake Adobe/Zoom updates to trick users into installing ConnectWise ScreenConnect, a trusted RMM tool that bypasses security systems. The backdoor access looks like normal IT support, giving attackers persistent remote control.
An npm worm in keyv@6.0.0 propagated through the dependency graph to infect 353-868 packages, specifically targeting Claude Code and VS Code to steal credentials and cloud tokens. The attack exploited keyv's high-trust infrastructure position and its transitive reach across production Node.js system
Over 24,000 exposed data centers still run CVE-2013-4786, a 22-year-old IPMI vulnerability that leaks password hashes and grants pre-OS access to baseboard management controllers. The flaw reflects industry-wide failure to patch its most privileged infrastructure.
A researcher demonstrated how to exploit Google's AI code review bot by bypassing privilege boundaries to steal GitHub tokens and gain system access, highlighting how misconfigured AI agents in developer infrastructure pose insider threats.
TP-Link's Omada ZTP has 15 interconnected architectural vulnerabilities—hardcoded keys, weak validation, and a race condition in cloud adoption—that allow unauthenticated attackers to intercept credentials and gain administrative control of the entire managed network.
DOUBLECUP malware hides JavaScript in steganographic PNGs cached by browsers, then executes payloads in-memory via fake login commands—evading endpoint defenses by leaving no disk artifacts.
CVE-2026-58048 in cPanel's database-renaming routine lets authenticated customers execute SQL with root privileges, breaching the trust boundary between hosting accounts and infrastructure. CVSS 9.4.
Madera Hospital took 14 months to notify patients of a May 2025 breach—far exceeding HIPAA's 60-day requirement. The "data review" excuse is increasingly used as a liability tactic to delay notifications.
Device code phishing (up 1,500%) exploits OAuth's device flow, tricking users into approving attacker sessions on Microsoft.com. Traditional defenses fail because the URL and certificate are legitimate.
N-central's patch bypass CVE-2026-18577 enables confirmed attacks on MSP infrastructure managing thousands of customer endpoints. CISA warns the supply-chain risk mirrors Kaseya VSA 2021, threatening MSPs and their entire client roster through a single compromised platform.
Microsoft's record $20M bug bounty program masks shifting priorities: new focus on supply chain security ($800k) and live hacking events ($2.3M), though AI is flooding submissions with both benefits and challenges.
Iran-linked attackers hit 30+ water systems in 7+ states. New York's $9M security grant ($59k per utility) exposes chronic underfunding of water infrastructure cybersecurity.