The day's cybersecurity headlines from public sources, in one place. Search them, filter by theme or CVE, or show only vulnerabilities CISA has confirmed are being exploited. Every headline links to the original.
For my own read on what the news means for your business, see Insights.
-
Unit 42 finds AI speeds malware development but does not improve success rates
Palo Alto Networks Unit 42 analyzed 405 AI-linked malware samples and found only 12 reached production endpoints. The finding is that generative tooling compresses the build phase without meaningfully raising the odds that the resulting malware works against defended environments.
-
OpenAI bans Russian accounts running an influence operation through a fake Israeli think tank
OpenAI removed a cluster of accounts that used VPNs to bypass access restrictions and generate social posts and comments distributed on Substack, Telegram, X, Facebook, and LinkedIn. The operation promoted the International Burke Institute, a self-described expert community based in Israel, and…
-
NSA to host a reunion for former Tailored Access Operations personnel
NSA will welcome potentially hundreds of former TAO members to Fort Meade to mark the unit's rebranding and to recruit alumni back. The agency restored the Tailored Access Operations name in July after a 2016 reorganization had folded offensive operations into broader directorates. TAO is expected…
-
Meta agrees to $17 billion and sweeping product changes to settle state AG suit over youth harms
Meta settled days into a civil trial brought by attorneys general from nearly every US state and territory, which alleged the company knew and concealed findings that Facebook and Instagram are addictive and that it violated COPPA by relying on self-reported ages. California AG Rob Bonta called it…
-
LACMA discloses 2025 breach affecting customer and employee data
The Los Angeles County Museum of Art disclosed that a breach last year exposed customer and employee information. Details on scope and cause are thin at time of writing.
-
DOJ and FBI disrupt QScan and QTRouter, Chinese platforms used against NASA, the Federal Reserve, and the Senate
DOJ announced the disruption of two hacking platforms operated by a Chinese state-sponsored group tracked as QTFY, employed by Nanjing Xinjiuwei Network Technology Company. Named victims include NASA, the Federal Reserve, the Department of Energy, DOJ, HHS, NIH, and the US Senate. Lumen Black…
-
Boston Scientific cyberattack halts order processing and shipping, disclosed to SEC
Boston Scientific discovered a cybersecurity incident on Tuesday, August 25, and filed with the SEC saying it has affected access to certain operating systems and business applications, including the ability to process and ship customer orders. The company declined to say whether ransomware was…
-
Adobe and Nvidia publish multiple advisories, including critical fixes
Both vendors released several advisories addressing dozens of vulnerabilities across their product lines, some rated critical. Routine but worth flagging for asset owners running Nvidia stacks in AI or visualization workloads.
-
Chrome 152 ships with more than 300 vulnerability fixes
Google released Chrome 152 with over 300 security fixes. Most were found by Google's own AI-assisted discovery work, though researchers are still reporting high-value findings independently. Volume at this scale is now routine and is a direct consequence of automated bug discovery.
-
Iranian IRGC-linked Nimbus Manticore adds a TWOSTROKE-like backdoor and SSH tunneler
Supporting: The Record, "Iran-linked hackers expand infrastructure across Europe and Middle East" |
-
SLEEPWALKER backdoor sits inert until one crafted packet arrives, then runs its own bytecode
An independent researcher documented a previously unreported Windows backdoor: an unsigned 64-bit DLL of 59,904 bytes built to be side-loaded into ERAAgent.exe, the ESET Management Agent executable. It impersonates Microsoft's dpapi.dll, exporting the same seven data protection functions, and…
-
AnonyMousKIT rents AI voice agents that call theft victims posing as Apple Support to strip Activation Lock
SOCRadar's Threat Research Unit detailed a credit-metered phishing-as-a-service platform built to unlock stolen Apple devices. From a single victim record it drives lures across five channels, with email at 1.50 credits, SMS priced per sender ID, WhatsApp, a recorded voice call at 1 credit, and an…
-
Two unpatched Kaltura video player flaws allow file read and code execution, and CERT/CC cannot reach the vendor
CERT/CC disclosed CVE-2026-19913 and CVE-2026-19912, both stemming from unsafe deserialization in the mwEmbedLoader.php endpoint of Kaltura's mwEmbed HTML5 player library, also distributed as html5lib. Neither requires authentication or a Kaltura session token, and network access to the endpoint…
-
NovaCookies phishing kit rents Microsoft 365 session theft for $320 a month using real Docusign envelopes
Island documented an adversary-in-the-middle toolkit that proxies Microsoft 365 sign-ins and captures authenticated sessions, sold as a subscription service. Campaigns used genuine Docusign envelopes to carry the lure, and some clicks were routed through legitimate Microsoft or Google sign-in…
-
CISA says over 100 internet-exposed water systems were targeted in July, most via PLCs on cellular modems
CISA has for the first time quantified the summer wave of attacks on the Water and Wastewater Systems Sector, saying it observed malicious activity against more than 100 internet-exposed systems in July, commonly PLCs wired straight to a cellular modem. The activity is linked to Iranian threat…
-
Scan of 3.5 million hosts finds 28,000 exposed .git directories leaking credentials and financial data
Intruder scanned 3.5 million live hosts and found roughly 28,000 publicly reachable .git directories containing API keys, bank details, and internal employee records including disciplinary files. The team built a tool called gitreaper that inspects repository history in memory rather than cloning…
-
Alice raises 140 million dollars for AI model defenses and enterprise guardrails
AI trust, safety, and security company Alice announced a 140 million dollar funding round to build defenses for AI systems against emerging vulnerabilities and adversarial attacks.
-
WhatsApp adds support for multiple passkeys per account across iOS and Android
Meta announced that WhatsApp now supports registering multiple passkeys to a single account, so users with both an iOS and an Android device can sign in with phishing-resistant credentials on either. Meta says more than one billion people already use a passkey to log into WhatsApp. The update also…
-
CISA publishes side-by-side red team results from two organizations, one detected and one did not
CISA ran two simultaneous red team assessments using comparable tradecraft against two critical infrastructure organizations. Against Organization A the team gained access to multiple workstations, escalated to domain-level privileges, and moved laterally into specialized business systems and…
-
Treasury sanctions six Iranian MOIS operatives who hit US energy, defense, healthcare, and universities
Operation Economic Outcast designated nearly 60 Iran-linked entities, individuals, and vessels across nuclear, missile, oil, and cyber networks, and named digital assets as a sanctionable sector. Six Iranian Ministry of Intelligence and Security operatives were sanctioned for compromising US…
-
Hospital operator Nutex Health discloses data exfiltration in SEC 8-K
Nutex Health, a Nasdaq-listed operator of 28 facilities across 12 states with 2025 revenue of about 875 million dollars, disclosed that an unauthorized third party accessed and exfiltrated information from its servers. The company engaged outside forensics, activated its response plan, implemented…
-
Marimo notebook flaw runs an attacker-supplied MCP command before any cell executes
CVE-2026-75149 is a code injection issue in Marimo versions prior to 0.23.15, scored 8.7 on CVSS v4 and 8.8 on v3.1. A crafted notebook can specify an attacker-controlled MCP server command in its configuration, and opening that notebook in edit mode launches the command as a local subprocess…
-
One website visit can permanently poison the local AI model behind NVIDIA NemoClaw
NemoClaw, NVIDIA's open source reference stack for running agents such as OpenClaw, started Ollama with OLLAMAHOST=0.0.0.0:11434, binding the local model server to every network interface. Oasis Security showed that a malicious webpage plus DNS rebinding gives an attacker unauthenticated control…
-
E4del and PINHOLE RATs pull commands out of FTP server banners
Two previously unreported remote access trojans are using FTP welcome banners as dead drop resolvers, reading command-and-control instructions from the text string an FTP server returns on connection. SOCRadar says this is the first time the technique has been seen in the wild, though…
-
Twenty-four npm packages used as free hosting for ClickFix fake CAPTCHA pages
Researchers found a cluster of 24 npm packages being used not to infect developers but as free, reputable, CDN-backed storage for phishing content. Each package contains a single HTML page that renders a fake Cloudflare CAPTCHA and redirects into a ClickFix flow. Installing the package does…