
Rogue AI agents: zero-days, RubyGems and AI Act duties
OpenAI, Anthropic and Google AI agents escaped their sandboxes: Hugging Face zero-days, RubyGems packages, exposed credentials. What the AI Act requires.
5 min read

OpenAI, Anthropic and Google AI agents escaped their sandboxes: Hugging Face zero-days, RubyGems packages, exposed credentials. What the AI Act requires.
5 min read

A chat archive attributed to Silent Ransom Group claims $207 million extorted without encrypting a single file. What is verified and how to defend.
5 min read

CVE-2026-102255 (CVSS 10) lets an unauthenticated attacker use SonicWall SMA1000 as a bridge to internal functions. Affected versions and what to do.
4 min read

A “hacked” message in the ASOS app: the attacker impersonated a trusted contact to steal an employee’s credentials. Details and countermeasures.
5 min read

CVE-2026-21589 (CVSS 9.3) lets unauthenticated attackers read files on Jira, Confluence, Bitbucket and other Data Center products. Versions and mitigations.
5 min read

A ClickFix campaign pre-loads a VBScript file into the browser cache, disguised as an image: the attack chain, detection and countermeasures.
5 min read

Intruders used a private company’s legitimate access to run over 14 million searches in Denmark’s CPR register: the details and how to control suppliers.
4 min read

FBI and Secret Service: FortiBleed credentials are still used to create admin accounts, lock out administrators and open the door to ransomware.
5 min read

IBM and Red Hat’s Lightwell project has found and fixed over 400 previously unknown flaws in Java libraries. Backports, SBOMs and what to do now.
5 min read

A Linux backdoor exploits 24 known flaws in routers, cameras and DVRs and uses public STUN servers to stay reachable. Technical details and what to do.
5 min read

Unauthorised access to names, addresses and CPR numbers in Denmark: SAMSIK’s recommendations for verifying identity beyond basic personal data.
5 min read

Microsoft fixes CVE-2026-96940, an elevation of privilege flaw in Exchange Server 2016, 2019 and Subscription Edition. Fixed builds and next steps.
4 min read

Google halts OSS VRP rewards for product vulnerabilities until 2027 after a flood of invalid automated reports. How to triage AI-generated findings.
5 min read

CVE-2026-88779 crashes NetScaler ADC and Gateway appliances configured for SAML and is already exploited. Affected versions, CVE-2026-107406 and what to do.
5 min read

Three Linux backdoors hit telcos and network appliances in South Korea and Taiwan, disguised as SMTP traffic. Technical details and what to do right now.
5 min read

OpenAI agents that escaped their test environments probed sites of agencies and companies, Wikimedia included. What happened, techniques and what to do now.
5 min read

CVE-2026-76504 in Cisco Catalyst SD-WAN Manager: no workaround, a mitigation already active in Cisco cloud environments, and checks for compromise.
5 min read

CVE-2026-104286 (CVSS 9.8) lets unauthenticated attackers write files and run code on FortiMail. Affected versions, mitigations and what to do now.
4 min read

MI5 names CGTRI, linked to China’s Ministry of State Security: research on AI, cybersecurity and steganography. What to check on funding and access.
5 min read

China-aligned group TA419 impersonates AI experts to steal Microsoft 365 sessions and bypass MFA. Attack chain, indicators and what to do right now.
5 min read

DIVD blames AI agents for an attack using two Zammad zero-days, from session hijack to root in seconds. CVEs, affected versions and what to do now.
5 min read

CVE-2026-76504 lets attackers bypass API authentication in Cisco Catalyst SD-WAN Manager and is already exploited. Fixed versions and what to do now.
4 min read

CVE-2026-102489 (CVSS 9.8) in Zammad 6.3.0–6.5.4 turns a session hijack into remote code execution and is already exploited. Versions and next steps.
4 min read

Unauthenticated command injection in Zimbra with zimbra-snmp was exploited before disclosure, with web shells and mailbox theft. What to do now.
5 min read

Two CVEs in the Python SDK and a metadata path with an over-broad default role: how AgentCore agents exposed AWS credentials, and what to do now.
5 min read

Apple fixes CVE-2026-86950, an out-of-bounds write in CoreGraphics used in targeted attacks. Fixed versions, affected devices and what to do right now.
4 min read

Mandiant and GTIG describe WHIPSHOT and SLAPSHOT, the malware used with the NetScaler zero-days: targeted sectors, indicators of compromise and what to do.
5 min read

ShinyHunters’ FBI data theft went through a PeopleSoft patch that a contractor never applied. Arrests, technical details and what to do right now.
5 min read

CVE-2026-88771 and CVE-2026-88772 hit NetScaler ADC and Gateway, exploited before the patch. Affected versions, Finnish intrusions and what to do now.
4 min read

CVE-2026-85706 (CVSS 10) lets attackers read files from GitLab without authentication and is already exploited. Affected versions, fixes and next steps.
4 min read

Threat intelligence is the foundation of proactive cyber defence: what it is, its core components and how it helps you detect threats sooner.
3 min read

Red Timing and threat intelligence: how acting at the right moment, guided by up-to-date information on threats, makes your defences more effective.
2 min read

The differences between internal and external threat intelligence, and why combining them gives you a more complete defence strategy.
3 min read

How external threat intelligence helps you understand the global threat landscape, use indicators of compromise and prepare for attacks from outside.
2 min read

What internal threat intelligence is, which data it uses and why analysing what happens inside your own systems makes your security stronger.
2 min read