Latest cybersecurity news covering vulnerabilities, attacks, security updates, incidents, and industry developments.
3603 articles
A September 23, 2026 webinar by BleepingComputer and Material Security will examine Google Workspace breaches involving social engineering and malicious OAuth apps, and discuss early response decisions and useful security controls.
Spain’s AEPD was notified of an alleged AI-agent attack that reportedly accessed systems, changed personal data, and viewed financial documents. The agency has not verified the incident and urges stronger, faster defenses.
A security newsletter roundup covers stolen AI keys used in intrusions, attacks on PaperCut and Microsoft 365 users, exploited Cisco flaws, spyware, security product updates, and new EU reporting obligations.
Microsoft is investigating missing Copilot buttons in Classic Outlook after upgrades to build 20026.20182 and later. Enabling “Show Apps in Outlook” may help; Microsoft also linked Outlook crashes to Kaspersky Mail Checker.
US, UK and Dutch agencies say Iranian state actors use Chosen Brick to surveil dissidents, activists and journalists, delivering it through messaging platforms and disguising it as legitimate software or medical files.
Windows Server 2022 enters extended support on October 13, 2026, with security updates through October 14, 2031. Microsoft recommends planning an upgrade to Windows Server 2025; Azure Edition hotpatching continues until October 2027.
DeepZero is an open-source Python tool that analyzes Windows kernel drivers for potential vulnerabilities. Its maintainer says it has found multiple verified flaws in the Snappy Driver Installer corpus; some are still undergoing disclosure.
Volodymyr Zelensky appointed former police chief and interior minister Ihor Klymenko to lead Ukraine’s National Cybersecurity Coordination Center, which coordinates government responses to major cyberthreats.
British, U.S., and Dutch agencies warn that Iranian state-sponsored hackers use CHOSEN BRICK spyware, delivered through tailored messaging lures, to surveil people viewed as threats to the regime and potentially increase their physical risk.
Norwegian agencies are investigating Telenor over alleged customer-data sharing with Myanmar’s military and possible sanctions violations in its subsidiary sale. Police raided Telenor’s Oslo headquarters; the company says employees faced serious risks.
A report says OpenAI uses outsourced reviewers to assess real ChatGPT conversations for response style. Privacy filters may miss personal details, and reviewers may see memory summaries that reveal additional user context.
UltraViolet Cyber launched Equinox, which assesses SIEM and EDR detection and log-source coverage against MITRE ATT&CK and ATLAS. The company says a customer trial raised mapped coverage from 26.6% to 61.3% without more alerts.
Postman announced general availability of Passport, a standalone API security product that keeps credentials in customer environments and gives developers and AI agents governed access to approved APIs.
CERT-AGID reported a phishing campaign impersonating Italy’s tax agencies with a fake holiday-bonus application. It solicits identity documents, payslips, a selfie, and contact details that could enable fraud and identity theft.
Memory-safe languages are gaining industry adoption as a way to reduce memory-related vulnerabilities. AI may help migrate legacy code, but rewrites need human review, and memory safety does not prevent other classes of flaws.
CrowdStrike says Falcon Data Security will use compact language models on-device to classify sensitive content in real time, starting with Intel NPU-powered PCs. The approach aims to protect data without sending it off-device.
CISA officials say the CDM program should expand automation, connect agency data and improve risk management. They cite SolarWinds lessons and outline priorities for shared capabilities, outcome-focused buying and continuous improvement.
A Swiss court sentenced a Ukrainian developer linked to LockerGoga, MegaCortex and Nefilim attacks to 12 years and nine months in prison. The appealable judgment also bars him from Switzerland for ten years.
Black Lotus Labs found BambooToken, active since at least 2023, using MQTT for command-and-control on Windows and Linux. Researchers linked it to compromised enterprise systems but could not attribute it to a specific actor.
Microsoft agreed to school AI standards barring most use of student data for training, sale, advertising or product development. The agreement also sets limits on AI companions and requires audits and transparency; OpenAI and Anthropic are discussing similar pacts.
Vercel says a two-week, $1 million sandbox challenge produced 1,285 filings and found two Linux kernel networking flaws, but no reports demonstrated access to customer data. Fixes are under private review.
The article says Nvidia limits Claude to low-sensitivity work, Palantir requires a zero-data-retention commitment, and Booz Allen bars its use for proprietary cybersecurity tasks amid concerns about retained user logs.
Wiz and Google Security Operations have expanded their integration with a shared data model, Blue Agent threat-analysis findings in Google Security Operations, and real-time bidirectional syncing of threat and case details.
Microsoft says the September 2026 KB5002914 security update can silently disrupt copy and paste, autofill, and formula dragging in Excel 2016–2024. Microsoft is investigating; uninstalling the update has restored functionality for some users.
