ls ./archive --page 13
Microsoft attributes StormEncryptor, a new C++ locker, to the China-linked group. Initial access was likely CVE-2026-18577 in N-able N-central, which is itself a bypass of the earlier fix. CISA has both flagged as exploited.
Attackers took credentials for a DigitalOcean Spaces bucket that BdThemes' promotional banner fetches JSON from. The plugin source stayed clean. Every admin page load ran the attacker's JavaScript anyway.
51 victims since April 2025, mostly outside North America, exploiting a 2024 Schneider Electric flaw and a 2025 Fortinet one. The advisory is worth reading for what it says about which patches actually matter.
Presented at DEF CON 34, 'Plug And Pwn' chains Windows' own driver auto-install against itself — emulate a Sierra Wireless modem, then a Sony FeliCa reader, and a DLL lands in System32. There is a Remote Desktop variant too.
Varonis got it with a URL parameter that pre-fills the chat — patched, $6,000 bounty. PromptArmor got it with white text inside an uploaded PDF, disclosed on 23 May and still unfixed when they published on 5 August.
Unauthenticated SQL injection through the password-reset endpoint, every release from 1.58 onward affected, and confirmed victims already naming what was taken. It is tracked as a GitHub advisory, which means NVD-driven scanners will not see it.
Ballena Azul did not exist. The recruitment did. Three developers came through a referral chain, signed contracts, got VM access — and within minutes ran dxdiag and checked which country their connection appeared to come from.
No external compromise, no evidence of misuse, and the key still had until March 2027 to run. Mozilla revoked it anyway — which is the right call, and a useful reference point for how a signing-key incident should read.
Black Hat and the weeks around it produced three unrelated results against phishing-resistant authentication: replayable signatures in Windows event logs, a master key pulled from Chrome's memory, and Windows Hello keys used without a PIN prompt.
ASSET Research Group's GhostSplice technique fragments an exfiltration request across MCP tool descriptions and tool results. No single piece looks wrong. The agent assembles it. Several models went from 0% to 100%.
CERT Polska's report on a December 2025 intrusion at a Polish CHP plant describes what it calls the first real-world use of a private APN as an attack path. The controller they landed on still had its default admin password.
Researchers from the University of Birmingham and Fuzzware showed that a hostile SIM can drive a standardised command called RUN AT straight into the modem. Three of eighteen phones accepted it. So did six of eight Quectel modules, and an EV charger.
Between July 2025 and June 2026 Microsoft awarded more than $20 million across 2,531 eligible reports from 562 researchers. The averages are the interesting part, and so is what they say about who can actually make a living this way.
The UK dropped its first secret order in 2025 after American pushback. It issued a second. Apple has now filed a complaint with the Investigatory Powers Tribunal, and the underlying demand is unchanged: access to backups that are end-to-end encrypted by design.
A March 2026 Interlock intrusion used Volatility3 — a legitimate incident-response tool — to pull domain credentials, NTLM hashes and account information out of memory. The rest of the chain was a ClickFix lure and PowerShell.
Sonatype tracked a campaign flooding npm with 846 packages carrying a multi-stage JavaScript loader. It fingerprints the operating system, fetches a matching payload, and on Windows patches security functions before establishing persistence through a scheduled task.
Manifold Security found that an authenticated non-admin user of Odysseus could execute OS commands with the process's own privileges. What that account reached is the story: API keys to spend, a mailbox to send from, and SSH keys to the machines Odysseus manages.
Mitiga found prompt-exfiltration tradecraft in the wild — malicious natural-language instructions in .cursorrules and CLAUDE.md that make the agent collect prompts, tokens and environment variables and send them onward. Also: 1,230+ hardcoded API keys sitting in AI instruction files.
Datadog Security Labs found that trusting a project is enough. In Codex, a project-scoped MCP config starts an attacker-controlled process; in Claude Code, a project-controlled PATH gets a repository wrapper executed by Claude's own Git probes. Neither needs a model response or an approval prompt.
Reported almost everywhere as a $1 billion fine, the figure is €890 million: €460m for preferencing its own services in Search and €430m for restricting how developers steer users to cheaper deals. The consequence that matters is what happens after day 60.
The week of 18–24 July 2026 was read as a physical-AI story, and Atoms' $1.7 billion made that easy. Underneath it, Cathedral, Glow and Neo Security took $360m between them, which says more about where security capital is going than any sector total does.
Silent Push counts roughly three million dangling DNS records worldwide. In one exercise across 12,500 apex domains they isolated 16,000 dangling subdomains and automated the takeover of 4,000 of them. The worst outcome isn't defacement — it's session cookies.
Sophos tracked STAC4749 across dozens of North American organisations between February and June 2026. Attackers used external Teams accounts and IT-themed personas to talk their way into remote access, then deployed a modular backdoor and, in at least three cases, Chaos ransomware.
Attackers changed passwords and IP addresses on controllers that were already reachable from the internet, locking operators out and triggering boil-water notices. Forescout then found 4,407 exposed Rockwell PLCs worldwide — most of them on mobile carrier networks.