Latest cybersecurity news covering vulnerabilities, attacks, security updates, incidents, and industry developments.
3800 articles
SecurityWeek's roundup covers an ADT-confirmed intrusion and customer data exposure, a Scattered Spider-linked arrest, CISA advisories, a Cursor IDE code-execution flaw, BlueNoroff lures, and Qinglong vulnerabilities exploited to install a cryptominer.
Dataiku’s Kiji Privacy Proxy locally detects and masks PII in requests before forwarding them to external AI APIs, then restores original values in responses. It uses a quantized DistilBERT model with ONNX Runtime.
Google is shifting Chrome and Android bug bounty priorities toward actionable reports as AI-assisted submissions surge. Android maximum rewards rise for some flaws, while standard Chrome payouts fall; Google expects total 2026 rewards to increase.
CIS announced updated and new security configuration benchmarks covering Microsoft 365 and Windows, AWS, Cisco, Tencent Cloud, SUSE Linux Enterprise, and Sophos Firewall, with changes including new settings and automated assessment content.
Microsoft’s optional Windows 11 KB5083631 preview adds a batch-file processing mode, expands the phased Secure Boot certificate rollout, and may trigger BitLocker recovery on some Windows Server 2025 devices.
Two US cybersecurity professionals received four-year prison sentences after pleading guilty to participating in BlackCat/ALPHV ransomware attacks. A third participant has pleaded guilty and is awaiting sentencing.
Cisco’s Model Provenance Kit fingerprints AI models using metadata, tokenizer similarity and weight-level signals. Its compare and scan modes help identify shared lineage and find the closest match in Cisco’s fingerprint database.
Two former cybersecurity incident-response employees were sentenced to four years in prison after pleading guilty to conspiring as BlackCat affiliates from May to November 2023. Prosecutors said a Tampa medical-device maker paid $1.27 million.
Malicious PyPI releases of PyTorch Lightning 2.6.2 and 2.6.3 run credential-stealing code on import; intercom-client 7.0.4 was also compromised. PyPI quarantined the project, and the campaign is assessed to extend Mini Shai-Hulud.
Congress extended Section 702 for 45 days. Senate Intelligence leaders committed to seek declassification of a FISC opinion within 15 days amid reported compliance concerns and ongoing debate over reforms.
The FCC approved measures to strengthen telecom customer verification and address illegal robocalls, end blanket authorization for certain foreign carriers, and refuse recognition of overseas testing labs without U.S. reciprocity agreements.
Former incident responders Ryan Goldberg and Kevin Martin were sentenced to four years each after pleading guilty to 2023 ALPHV/BlackCat ransomware attacks against U.S. organizations; one victim paid $1.3 million. Co-conspirator Angelo Martino separately pleaded guilty.
Bluekit combines 40+ phishing templates with campaign management, victim-session monitoring, and an AI assistant. Varonis found the AI-generated drafts were experimental and contained placeholders.
Anthropic launched Claude Security in public beta for Enterprise customers. The tool scans selected code repositories for vulnerabilities, explains findings, and provides confidence ratings and targeted patch instructions.
This weekly roundup covers Checkmarx’s confirmed 96GB data dump linked to credentials from the TeamPCP-attributed Trivy compromise, Copy Fail Linux LPE, ShinyHunters’ claims involving Medtronic and Carnival, and security industry and insurance findings.
Reports and Acronis support information say Windows 11 update KB5083769 causes VSS timeouts that can fail third-party backups on versions 24H2 and 25H2. Uninstalling the update is a temporary workaround; Microsoft had not responded.
The FBI says cyber-enabled cargo theft losses in the U.S. and Canada approached $725 million in 2025. Criminals use phishing and impersonation to hijack shipments; the bureau urges secondary-channel verification and MFA.
Thomasz Szabo was sentenced to four years in federal prison for leading an online swatting ring that targeted public officials, journalists, and religious institutions. He pleaded guilty to conspiracy and threats involving explosives.
CrowdStrike says Cordial Spider and Snarky Spider use voice phishing and social engineering to access identity platforms, steal data across victims’ SaaS environments and extort U.S. organizations.
Wiz says its AI-powered Red Agent, now in public preview for eligible customers, scans over 150,000 production web apps and APIs weekly and identifies more than 3,000 high- and critical-risk attack chains per week.
Researchers say four SAP-related npm versions in the Mini Shai-Hulud campaign delivered an information stealer targeting developer and cloud credentials. The versions were exposed for 2–4 hours, then unpublished and replaced.
CIS and security partners introduced three companion guides extending the CIS Controls to AI systems, with the LLM guide addressing context integrity, sensitive data, deployment choices and model supply-chain security.
An international operation led by Dubai Police arrested at least 276 suspects and shut nine crypto investment fraud centers. U.S. authorities linked millions in losses to the schemes, which used romance-baiting to lure victims to fake platforms.
The episode examines a malicious Bitwarden CLI release, npm lifecycle-script risks, OWASP guidance, and reports of AI-assisted malware adapted across package ecosystems. The speakers say the Bitwarden version was removed and question the attackers’ Shai-Hulud claim.
