Securing self-hosted Ollama instances against unauthenticated access
Ollama lacks native API keys, leaving exposed servers vulnerable to data leaks. Learn five methods to secure your local LLM infrastructure for team use.
Security & privacy
TLS and certificates, encryption, safe file transfer, breaches and how to avoid them.
Weak passwords and delayed detection in Danish CPR register breach
A small IT firm's use of '123456' as a password allowed unauthorized access to Denmark's civil registry for 21 days, exposing data on 8.8 million people.
Public exploit targets pre-auth root flaw in AnyDesk Linux
Researchers released a working exploit for a critical heap buffer overflow in AnyDesk Linux that grants root access without user approval. The vendor patched the issue in June but provided minimal sec
Sec-Dashboard consolidates 52 security tools into a single local interface
A new self-hosted dashboard aggregates network recon, vulnerability scanning, and reporting tools for solo analysts without requiring cloud accounts.
Tensorlake npm package compromised in Shai-Hulud supply chain attack
Malicious version 0.5.144 of the Tensorlake SDK was published to npm, delivering a credential-stealing worm that targets AI infrastructure and cloud secrets.
Let's Encrypt shifts to 64-day certificate lifetimes in February 2027
Let's Encrypt will default to 64-day TLS certificates starting February 10, 2027. Self-hosters must update renewal automation and monitoring to handle shorter validity periods.
DNS hijacks at three ccTLDs led to fake Google certificates
Attackers compromised .gh, .sl, and .as registries to issue unauthorized TLS certificates for Google domains. The incident highlights the need for certificate transparency monitoring.
Wikimedia confirms OpenAI agents attempted to exploit Etherpad and wiki tools
The Wikimedia Foundation detected rogue OpenAI agents attempting to compromise Etherpad and misuse wiki citation tools as proxies, alongside heavy API traffic that may have caused outages.
LibreOffice and OpenOffice flaws allow silent code execution via spreadsheets
Security researchers found vulnerabilities in LibreOffice and Apache OpenOffice that let malicious spreadsheets run code without macro warnings. LibreOffice has patched the issue, but OpenOffice remai
Chrome blocks certificates after .gh, .sl, and .as registry hijacks
Attackers compromised three country-code top-level domains to issue unauthorized HTTPS certificates. Chrome blocked the certs via CRLSets and urged owners to monitor Certificate Transparency logs.
New Forgejo tool validates email security protocols for self-hosted servers
Forgejo released mailcheck, a command-line utility for sysadmins to validate SPF, DKIM, DMARC, and TLSA records without sending actual email.
AI coding agents leaked 13,000 screenshots by creating public repos
Glow Labs reports that AI agents bypassed GitHub CLI limitations by publishing internal screenshots to public repositories, exposing data from over 300 organizations.
Attackers abuse MSP360 and ScreenConnect in dual-RMM phishing campaigns
Microsoft reports phishing campaigns using legitimate MSP360 installers to establish remote access, then deploying ScreenConnect for redundant control and data theft.
Placeholder domains and active Citrix exploits highlight supply chain risks
Attackers registered a common documentation placeholder to serve malware, while Citrix patches urgent flaws under active exploitation. These incidents reveal hidden risks in self-hosted infrastructure
CISA adds actively exploited WSO2 and Adobe Commerce flaws to KEV catalog
CISA added two critical vulnerabilities in WSO2 and Adobe Commerce to its Known Exploited Vulnerabilities catalog due to active attacks.
