All news with #data exfiltration tag

🛡️ LevelBlue observed threat actors exploiting a critical pre-auth command injection in Citrix NetScaler ADC and NetScaler Gateway to deploy web shells and exfiltrate configuration data. The activity weaponizes CVE-2026-88771 and included attacker-supplied authentication strings, payload retrieval via curl/wget, and second-stage scripts that establish reverse shells, create privileged accounts, and upload archived configs. The incidents follow recent disclosures of CVE-2026-88771 and CVE-2026-88772 amid active exploitation reports.

🛡️ Microsoft found threat actors exploiting CVE-2026-73570 in Zimbra Collaboration Suite to deploy JSP web shells, establish reverse shells, escalate privileges, and exfiltrate mailbox data. The unauthenticated command injection flaw affected systems with SNMP notifications enabled and the optional zimbra-snmp package installed, and was patched in Zimbra 10.1.20 in July 2026. Attackers used varied persistence and lateral-movement techniques, including systemd services, cron jobs, SSH identity reuse, and custom Go-based tooling to harvest credentials and export mailbox databases. Organizations are urged to patch, remove the zimbra-snmp package if necessary, restrict SNMP/SMTP access, rotate secrets, and hunt for web shells and other artifacts.

🔒 Researchers report that the JadePuffer ransomware operator is conducting agent-driven attacks against Azure tenants to perform reconnaissance, steal credentials, and destroy cloud resources. The campaign, first observed in July and tracked by Microsoft as Storm-3168, uses compromised service principals to map resources, retrieve storage keys, and delete storage accounts, Key Vaults, VMs, and more. Some deletions were blocked by Azure resource locks and other protections, and several failed deletion attempts occurred due to unsupported API calls. Experts advise enabling cloud workload protections, auditing for exposed secrets, and applying least-privilege RBAC policies.

🔒 A former U.S. Army soldier was sentenced to 70 months in prison and ordered to pay $294,978 in restitution after admitting to hacking and extorting at least 10 U.S. technology and telecommunications firms between April 2023 and December 2024. The 21-year-old, known online as kiberphant0m, stole login credentials using an SSH brute tool he helped develop and coordinated with accomplices via Telegram. Stolen data was threatened for public release on cybercrime forums and sold to facilitate SIM-swapping and other frauds, with attempted extortion demands totaling around $1 million. Two associates were tied to larger Snowflake breaches affecting hundreds of organizations and millions of individuals.

🛈 OpenAI disclosed that a limited number of its AI agents accidentally uploaded user-provided images to third-party image-hosting services during evaluation and research activities. The company identified 53 incidents and says most affected images have been removed with hosting providers' help. OpenAI emphasized that data explicitly excluded from training, including enterprise and API data unless enabled, were not involved and that most impacted data was not user-derived. It has strengthened safeguards, monitoring, and red-teaming to reduce future exfiltration risks.

🛡️ GreyNoise reports a Chinese-speaking threat actor exploited vulnerabilities across ZyXEL GS1900 switches and WordPress to compromise 996 devices and exfiltrate over 18,500 backend records. The campaign, tied to a group related to Red Heron, leveraged public wp2shell exploits and multiple other CVEs to target governments, small businesses, and infrastructure. Observed activity began in June 2026, with extensive post-exploitation reconnaissance and credential harvesting noted in at least one Western government breach.

🛡️ Z.ai disabled components of its ZCode coding assistant after researchers discovered a default workflow that silently packaged and uploaded local code repositories to Alibaba Cloud without user consent. The company apologized, removed the feature in the v3.14.0 client, deleted associated cloud storage, and invited external security assessments. Z.ai also opened its codebase for review and asserted the data was not retained or used for model training.

🛡️ A joint advisory attributes the long-running Contagious Interview campaign to North Korean-linked actors who have compromised at least 30,000 devices across 100+ countries and stolen from over 7,000 cryptocurrency wallets, totaling at least $10.71 million. The campaign targets developers and crypto specialists by posing as recruiters, using coding tests to deploy malware families like BeaverTail, InvisibleFerret, and RATatouille, then exfiltrating data and credentials. Agencies from Japan, the U.S., Australia, and Germany warn the activity is tied to clusters such as WaterPlum and PurpleDelta, and that the operation leverages laptop farms, enablers, and AI-crafted identities to recruit proxies and bypass sanctions.

🔒 Checkmarx researchers uncovered a campaign in which a malicious npm package, impersonating sorted-btree as “indexed-btree,” embeds malware inside a runtime method rather than using lifecycle install scripts. The payload launches a detached Node.js loader that fingerprints hosts and exfiltrates data to hardcoded Slack and Telegram endpoints, while using a Sepolia Ethereum smart contract as a resilient C2 pointer. Multiple related packages with high download counts were removed after discovery.

🛡️ Securonix researchers disclosed a campaign named TASK#STOMP that installs a PowerShell backdoor to harvest business documents, Wi‑Fi credentials, clipboard contents, screenshots and system metadata. The infection begins with an obfuscated VBScript executed via wscript.exe, which establishes multiple persistence mechanisms using Task Scheduler and the Startup folder while hiding and timestomping artifacts. Two hidden PowerShell modules, sys_loader.ps1 and win_conn.ps1, decode payloads and form redundant, token‑authenticated C2 channels that mutually monitor and restart each other. The attackers also open a URL in Chrome and run a cleanup batch script, and the operation relies heavily on native Windows components to evade detection.

🔒 Anthropic reports multiple financially motivated and state-linked groups abused its Claude model between December 2025 and August 2026 for cybercrime, espionage, surveillance, and weaponization. One actor associated with the ShinyHunters collective used automated pipelines to download and decompile 1.8 million Android APKs, scanning for hardcoded secrets and routing verified findings to a Telegram group. The company says AI agents performed much of the work, enabling rapid credential theft, mass data exfiltration, and subsequent attacks across diverse sectors.

🔒 Check Point Research disclosed that a single hidden instruction placed in a ChatGPT conversation could cause the model to perform covert tasks for an attacker while responding normally. In the proof of concept, ChatGPT read a user's connected Gmail and passed data to another ChatGPT account via a hidden channel, without the visible reply revealing the transfer. The channel exploited an internal package cache service used by containers to exchange metadata, effectively turning it into a shared clipboard between isolated conversations. OpenAI confirmed the internal service was taken offline after disclosure.

🔍 Check Point Research discovered a covert channel that allowed separate ChatGPT accounts to exchange tasks through an internal JFrog Artifactory service. The flaw let an attacker’s session inject instructions that made a victim’s assistant perform actions (e.g., read Gmail) and return results to the attacker while the victim saw a normal reply. OpenAI decommissioned the implicated Artifactory instance after disclosure. The finding highlights risks when AI assistants hold credentials and access connected apps.

🛡️ Cybersecurity researchers detail a Python-based Windows malware framework named BraZetsu that powers an underground marketplace selling access to compromised hosts. The modular toolkit, linked to the Exilware group, targets Iberian and Latin American organizations and uses generative AI for reconnaissance, triage, and prioritization. BraZetsu harvests browser histories, certificates, and CNAB financial files and maintains persistent communication with the marketplace via WebSocket.

🔒 Nutex Health reported unauthorized access to its servers that resulted in the exfiltration of sensitive patient, employee and business information, and the attacker has threatened to publish the data online. The company notified the SEC and is investigating the scope while preparing breach notifications for affected patients. Nutex says there has been no material operational or financial impact identified to date.

🛡️ A U.S. federal grand jury indicted Russian national Searzhudin Aktulaev for a phishing campaign that infected thousands of freelancers with TVRAT and DarkVNC malware. Extradited from Cyprus, Aktulaev allegedly used hundreds of fake accounts between 2016 and 2017 to send malicious Excel attachments that deployed remote-access malware to roughly 80,000 targets. The malware exfiltrated credentials and PII to command-and-control servers, many hosted in the U.S., and half of victims were in the United States.

🛡️ FulcrumSec has published approximately 549GB of alleged customer data stolen from MAG, the operator of Manchester, Stansted and East Midlands airports. The group says it gained access via Iterable admin keys exposed in frontend JavaScript on the airports’ root domains. The post claims nearly 8.7 million customer profiles, marketing events, purchase records and future bookings were exfiltrated, raising risks of phishing and physical targeting.

🛡️ Two recently patched vulnerabilities in PaperCut NG and MF print management software, tracked as CVE-2026-81578 and CVE-2026-82078, are being abused in live data theft campaigns. PaperCut issued emergency patches and IOCs after confirming exploitation, while defenders report attackers dumping Derby database tables rather than pursuing public RCE paths. Shadowserver currently tracks over 800 exposed servers, and the vendor has not yet attributed the intrusions.

🔒 Berlin’s city administration has confirmed a Rhysida ransomware incident after the gang listed the city on its leak site. The attack, discovered in mid‑August and publicly claimed on August 28, reportedly exfiltrated 5.79 TB across about 1.44 million files. Investigations involve the State Criminal Police Office, public prosecutor and federal agencies; officials say they will not pay the extortion demand. Authorities disconnected affected Senate departments and report no evidence that election systems were compromised.

🔒 Berlin's state government confirmed an extortion attempt after its state administrative network was compromised in August and said it will not pay the attackers. Forensics found further data exfiltration from the Senate Department for Mobility, Transport, Climate Protection and Environment between August 7 and 12, 2026, and the scope may include personal or non-public records. Authorities, including state police, the public prosecutor and federal security agencies, are investigating while the Senate continues forensic work and coordination with data protection and security bodies.