Breaches, vulnerabilities and the people cleaning up after them. Reporting for defenders, not for panic.
Raheim Hamilton, who ran one of the largest dark web marketplaces, received a four-decade prison sentence for overseeing more than 4 million illegal transactions facilitated through cryptocurrency between 2018 and 2020.
Law enforcement warns that attackers continue exploiting a massive Fortinet credentials leak to lock out legitimate administrators from FortiGate firewalls and SSL VPN gateways, with ransomware groups leveraging the compromised access.
Threat actors compromised operators of country-code registries for Ghana, American Samoa, and Sierra Leone, then modified DNS records to obtain unauthorized HTTPS certificates for Google domains and other organizations.
David Robinson, who recently departed OpenAI, describes a shift from quarterly model releases to weekly capability and risk updates driven by reasoning training, tool integrations, and AI-assisted research.
As AI agents write an increasing share of code on GitHub, secret leaks are accelerating faster than developer carelessness alone would explain. GitHub and Microsoft have built a new classifier to catch more credential exposures before they reach public repositories.
Cloudflare has built a multi-AI-agent system to handle the overwhelming volume of security alerts that arrive simultaneously, using specialist agents to gather evidence, connect detections, and recommend actions while keeping human analysts in control of final decisions.
Microsoft is expanding its attachment security controls by blocking .msix and .msixbundle files in Outlook Web and the new Windows client, part of a broader campaign to prevent attackers from exploiting installation package formats.
Researchers tracking a cryptomining campaign have identified PoeLLM malware infecting over 3,400 servers running exposed AI tools. The botnet uses an unusual technique to retrieve command-and-control addresses by parsing keywords from a poem stored on GitHub.
