World watch

Microsoft adds task-based access and risk scoring for AI agents after treating them as insiders in a new control set now reported.

A Russian espionage group carried out a two-month intrusion at a naval shipbuilder in two NATO countries to steal credentials and tunnel into the network.

UK finance watchdog asks firms to report AI failures and close calls to assign clear accountability in financial services.

Cyber resilience is set as a boardroom priority for UK public services according to reporting from The Register.

Kaspersky experts report initial access vectors and compromise impact for warehouse environments in a new technical analysis.

Anubis listed Fisher59 on its leak site as a victim, with the report stating a data breach at a beverage distributor from one outlet.

Malware News reports a scheduled ISC Stormcast episode with a noted course offer for binary triage skills training now open for enrollment.

ICANN is testing single sign-on technology to give law enforcement access to non-public WHOIS data in cases of imminent threat.

Ransomware groups published claims against six entities in the past 18 hours from one reported source.

The report comes from Dark Reading. 1 outlet reported this event inside 18 hours. Here is what happened, why it matters, and what to do next.

Ransomware operators exploit PAN-OS CVE-2026-0257 in reported attacks on network security systems.

Researchers infiltrated a large Microsoft tech support scam run by a Russian organization using a malicious TDS and abusing Cloudflare links.

Threat actor used ConnectWise Cloud Backup API secret for unauthorized Microsoft 365 access; ConnectWise disabled backups and impact remains unconfirmed.

Ahold Delhaize confirmed a cyberattack forced US and European systems offline, marking the grocery giant's second security incident in two years.

Cloudflare cut AI inference costs by 68 percent using prompt caching and model routing after per run costs doubled.

Senator Ron Wyden has called on hundreds of federal agencies to implement encryption technologies to protect the sensitive online traffic of visitors to...

Cloudflare introduced Clairic, a new System One class of extraction models, on 2026-10-19. The source gives no further details.

DHS will ban DeepSeek from federal civilian agencies over national security concerns, including data sent to servers in China and potential malware.

Drupal warns of fast exploitation risk for a critical core RCE flaw affecting Entity List and Field UI modules.

Microsoft adds new capabilities in Microsoft 365 E7 and Project Perception to contain agent risk as agents join daily work across business systems.