Apple’s Spyware Pager Goes Off in 110 Countries

Apple confirmed to TechCrunch it sent a fresh round of mercenary-spyware threat notifications on August 13, reaching users in 110 countries — part of a running total the company now puts at over 150 countries notified since the program started in 2021. The alerts, now surfaced directly on the iPhone lock screen in addition to email, read plainly: a mercenary spyware attack was detected targeting the device. Apple won’t attribute the campaign to any vendor or government, though it has previously named NSO Group’s Pegasus as the archetype of the category, and it maintains it can never achieve absolute certainty — only high confidence.

This is the surveillance-industry equivalent of a smoke detector going off in 110 buildings at once, and Apple still won’t say who’s holding the lighter. Journalists, diplomats, activists and politicians remain the target profile, but the scale of this round is a reminder that mercenary spyware isn’t a niche threat model anymore — it’s a recurring, global, multi-vendor business. If you receive one of these notifications, take it at face value and get expert help; if you don’t, the operative word in “vast majority of users” is still majority, not all.

More via The Hacker News 

GeoServer’s Unpatched Zero-Day Is Getting Probed Right Now, and There’s No Fix Yet

A researcher going by @q1uf3ng dropped an unauthenticated SQL-injection zero-day in GeoServer’s jsonArrayContains function on X on August 12, complete with a working path to remote code execution against admin-configured databases. WatchTowr says exploitation attempts started within hours — hundreds of probes from a small pool of IPs, currently reconnaissance rather than payload delivery, with the RondoDox botnet already in the mix. Shadowserver counts over 1,500 exposed instances worldwide. As of this writing there’s no CVE, no CVSS score, and no vendor patch.

GeoServer sits quietly underneath a lot of government, agriculture, telecom and transit infrastructure that nobody thinks about until it’s the reason a map doesn’t load — or the reason an attacker has a shell. No patch means your only lever right now is exposure reduction: get it off the public internet, restrict jsonArrayContains where you can, and watch your logs for the probing pattern, because “reconnaissance now” reliably turns into “exploitation later” once someone weaponizes the PoC properly.

More via The Hacker News 

The March Supply-Chain Attack That Keeps Getting Bigger: LiteLLM’s 2,500-Company Hangover

Back in March, threat actor TeamPCP compromised Aqua Security’s Trivy scanner, which LiteLLM’s own CI pipeline auto-installed, which let attackers slip credential-stealing code into two LiteLLM package releases that sat live on PyPI for about 40 minutes before being pulled. New victim-mapping published this month by CloudSEK puts the real damage at over 2,500 organizations and roughly 430,000 exposed CI/CD pipelines — cloud keys, Git tokens, Kubernetes secrets and AI-provider API keys, scooped up across a supply chain most of those victims didn’t know they were part of.

LiteLLM is the plumbing that routes requests to over a hundred LLM providers, which means this wasn’t an attack on LiteLLM’s users so much as an attack on everyone downstream of everyone who ever ran it. Five months on, boards that closed the ticket after “we unpinned the bad version” are now finding out that credential rotation, not version pinning, was always the actual fix. If your CI/CD ever touched LiteLLM 1.82.7 or 1.82.8, or anything built on top of Trivy in that window, that’s your homework this week.

More via SecurityWeek 

A Nation-State Crew Rooted 361 vCenter Servers in 47 Countries — Patch Was Out Five Days

Broadcom patched two critical VMware vCenter flaws on July 29 — a directory-traversal bug and an auth-bypass, both rated a maximum 9.8. Five days later, a suspected China-nexus actor was already inside. Researchers at German incident-response firm QUIRSO tracked the campaign from first contact on August 3 through August 9, by which point they’d counted 361 distinct victim IPs spread across 47 countries, with Germany, the US, Turkey, Iran and France taking the brunt of it. The attacker used a reverse SSH shell for persistence and, on at least one appliance, chained both flaws together to spin up a rogue admin account with no corresponding login from the legitimate one. Some infections have since been tied to Babuk-derived ransomware.

vCenter isn’t just another box on the network — it’s the console that owns every VM, every host, every snapshot in an organization’s virtual estate. Compromise it and you don’t have a foothold, you have the building. The five-day gap between “patch available” and “hundreds of servers owned” is the real story here, and it’s not an outlier anymore, it’s the baseline. If you’re running vCenter and it’s reachable from anywhere near the internet, assume compromise-until-proven-clean, not patch-and-move-on.

More via The Hacker News 



Nine Fortune 500 Names, Millions of Records: The Azure Credential Dump Nobody’s Calling a Breach

A threat actor calling itself “TheHatman” has spent the past week flooding cybercrime forums with internal employee directories allegedly pulled straight from the Azure and Entra tenants of nine major companies: McDonald’s (1.7 million records, the largest haul), Tata Consultancy Services, Vodafone, HCL Technologies, IHG, Kyndryl, Gap, Hexaware and Wyndham. Hudson Rock says the data — names, roles, emails, phone numbers, workplace addresses — matches standard Azure directory export formats closely enough to look legitimate. TCS has told its stock exchange it’s found no credible evidence of a breach on its own systems.

Researchers are converging on compromised credentials and infostealer malware rather than an Azure platform vulnerability, which is almost worse news: it means this isn’t a patchable bug, it’s the modern reality that your perimeter is whatever your weakest employee’s browser has saved a password to. Nine unrelated global enterprises hit by the same actor via the same access pattern is a services-and-identity story, not a Microsoft story — audit your conditional access policies and rotate anything infostealer logs might have touched, because “no evidence of a breach” and “not compromised” are not the same sentence.

More via SecurityWeek