Developers are combining frontier AI models with NVIDIA Omniverse libraries to assemble simulation applications. The workflow spans assets, physics, rendering and checks that scenes behave as intended, supporting scenario exploration and design investigations.
Anthropic is bringing its frontier models and onsite engineers to security companies protecting power grids, water systems and other critical infrastructure. Its accompanying OSS Scanner offers opt-in open-source projects periodic vulnerability scans at no cost.
CISA added five vulnerabilities to its Known Exploited Vulnerabilities catalog after reported exploitation by the China-linked Flax Typhoon group. The report identifies an October 11 remediation deadline for federal agencies and includes a critical improper-access-control flaw in ProFTPD among the affected vulnerabilities.
BleepingComputer reports active exploitation of CVE-2026-102255 in SonicWall SMA1000 appliances. The maximum-severity vulnerability already has a patch, making the reported attacks an urgent item for teams operating affected appliances.
TechCrunch reports that OpenAI's submitted mathematical proofs deviated from guidelines set by researchers the lab had consulted. The dispute highlights the importance of evaluating AI-generated proofs against the standards of the research community receiving them.
Google is turning Gemini into an agent that can plan and execute tasks across business applications and systems. The reported capabilities include delegating to subagents, using multiple AI models and operating with a workplace identity that includes an email address.