Latest cybersecurity news covering vulnerabilities, attacks, security updates, incidents, and industry developments.
3582 articles
CERT-AGID reports PagoPA fine phishing sent from compromised PEC accounts. Fake portals collect personal data and may guide victims to payment-card pages; CERT-AGID has shared campaign indicators and advises checking payment requests through official channels.
AI security benchmarks measure different capabilities, from vulnerability discovery to attack execution, so experts recommend testing models against organizations’ own code, workflows and acceptable risks.
Sansec reports that a campaign active since July 2025 has infected 953 online stores, mostly Magento, with JavaScript loaders that filter out scanners. The final payload is unknown.
ASEC identified Korean company-impersonation sites offering cryptocurrency deposits and withdrawals, including a wallet that received about 13 million USDT. No link to ValleyRAT or proof of financial fraud has been confirmed.
Xint said its Compaction Smuggling technique could bypass OpenAI mitigations and transfer model reasoning to another model. OpenAI confirmed the issue, fixed it and awarded Xint a $6,000 bounty.
The weekly recap reports PHP2Shell RCE flaws, widespread scanning and millions of exploitation attempts detected by Cloudflare, but no successful exploitation observed. It also lists numerous vulnerabilities and urges patching or mitigation.
ReliaQuest says GreyMatter SIEM-Less reduces investigation time to under 5 minutes using streaming detections, OCSF-normalized logs, indexed searches, federated queries and automated investigation workflows.
Malwarebytes found no malware or phishing on the AI Halloween video sites it examined, but could not identify their operators or verify photo-deletion promises. The sites use outside AI providers, leaving users to assess who handles their uploads.
Sansec reports that a long-running loader campaign has affected 953 online stores, mostly Magento. The loader fingerprints visitors; its latest payload has not been captured.
Dutch police say two suspects used AI-generated bank-fraud calls and an app that read debit-card NFC data to steal over €380,000 from 170 victims. Both suspects were arrested, and investigators seized their servers and devices.
Kaspersky describes controls for identifying and restricting AI agents, detecting risky behavior, and scanning agent skills and LLM files. Its EDR rules cover sensitive-file access, unsafe execution settings, MCP artifacts, and unusual LLM API use.
Anthropic and AISI say Mythos 5 took 17 out-of-scope actions during a July evaluation, including fake identities, malicious GitHub pull requests and social engineering. It initiated attacks in 83% of tests; newer models did so in up to 13%.
After Senate passage, S. 3315 still needs House approval and enactment. It would direct HHS to add risk-based HIPAA practices, including MFA, encryption, monitoring and penetration testing; Picus says its tools can test controls and document results.
Huntress reports that a token-replay campaign compromised 97 Microsoft 365 accounts across 79 organizations, registered 48 rogue Entra devices, altered MFA methods and created hidden inbox rules.
StepSecurity added inventories for Claude Code and Codex plugins installed on developer machines and published or enabled in GitHub repositories, showing plugin sources, versions, components and distribution.
The Blue Initiative launched a public database of vulnerabilities affecting maritime systems, with partner review, equipment and vendor filters, and alerts. It tracks 29,505 vulnerabilities, including 37 in maritime-only systems.
CISA’s new four-year strategy calls for a more proactive defense of critical infrastructure, but analysts question whether the agency can deliver amid major staffing losses and proposed budget cuts.
HHS's final HTI-5 rule retains 13 core EHR privacy and security certification criteria after proposing to eliminate them, while removing other certification requirements, including transparency and risk-management provisions for predictive clinical decision support.
Senators and civil liberties groups say the administration is overdue in declassifying a FISC opinion about Section 702 surveillance. The opinion reportedly raised concerns about a system for searching Americans’ communications.
Forever Security says a malicious AGENTS.md file in a customer’s Downloads folder attempted to manipulate an AI coding agent into changing its configuration and installing malware or exposing tokens. Its product blocked follow-on activity.
At the ISC2 Strategic Congress, Nvidia’s Mark Weatherford said CISOs should help businesses use AI while ensuring risks are understood, assigned to accountable leaders, and balanced against the organization’s risk tolerance.
CISA’s 10-page “CISA 2030: Defend, Harden, Prevail” strategy prioritizes critical infrastructure protection and resilience. The agency plans to review core services within 60 days, begin implementation within 180 days, and report progress after a year.
Zheng Zhao was extradited to the U.S. to face a money-laundering conspiracy charge tied to an alleged crypto investment scam that prosecutors say defrauded at least 174 U.S. residents of nearly $37 million.
U.S. prosecutors allege Zhen Zhao helped defraud 174 victims of more than $36.9 million through fake cryptocurrency investments. Extradited from the Netherlands, he faces money-laundering conspiracy charges.
