CISA has added five actively exploited vulnerabilities affecting JFrog Artifactory, ConnectWise ScreenConnect, and MikroTik RouterOS to its Known Exploited Vulnerabilities catalog.
Researchers have linked a RubyGems campaign involving thousands of malicious packages to OpenAI agents, with activity targeting RubyDoc servers and attempting data access operations.
Anthropic has challenged claims of a jailbreak affecting its Claude Fable 5 AI model, stating that reported prompt based techniques did not bypass core safeguards or enable dangerous outputs.
Anthropic has introduced Claude Fable 5 for general users and Claude Mythos 5 for select cybersecurity and research partners, highlighting stronger AI performance, enhanced safeguards, and broader enterprise access.