{
  "id": 145743,
  "title": "Open Secure AI Alliance Expands at Black Hat: What You Should Know",
  "url": "https://urgent.news/2026/08/04/open-secure-ai-alliance-expands-at-black-hat-what-you-should-know",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-04T18:11:27.000Z",
  "source": {
    "name": "TechRepublic",
    "slug": "techrepublic",
    "url": "https://www.techrepublic.com/article/news-open-secure-ai-alliance-safe-guidelines-black-hat/"
  },
  "original_language": "en",
  "account": "The Open Secure AI Alliance (OSAA) announced significant expansions at the Black Hat USA 2026 conference, introducing new guidelines and open tools to enhance enterprise AI security. The alliance, focused on making AI defense open, inspectable, and enterprise-ready, focuses on creating a robust ecosystem where IT leaders can deploy safer AI solutions.\n\nOSAA unveiled the Shared AI Findings Exchange (SAFE) guidelines, which aim to transform cybersecurity incidents and near-misses into shared protection across enterprises. Instead of relying on non-disclosure agreements or single-vendor ecosystems, SAFE provides a framework for turning breach details into collective defense strategies. This initiative allows organizations to share information about AI-related security incidents, promoting a collaborative approach to threat mitigation.\n\nA key component of the OSAA's expansion is the introduction of open agent-security tools. These tools provide IT leaders with inspectable frameworks and lower token costs, enabling enterprises to deploy safer AI solutions. By fostering an environment where AI defense stacks can be built using open standards and components, OSAA aims to create a more interoperable and scalable ecosystem for AI security.\n\nNvidia played a crucial role in the alliance's expansion by backing the OSAA through its Nvidia Labs Object-Oriented Agent (NOOA) framework. NOOA offers open agent-harness research, which shifts the focus from model choice to harness architecture. This approach allows agents to be developed as traditional software, making them testable, auditable, and governable by security and platform teams. The harness capabilities, outlined by NOOA, significantly impact model performance and token consumption, making them crucial for enterprises navigating the challenges of LLM costs.\n\nCisco's participation in the alliance further strengthens its impact, as Cisco integrates open harnesses and models into its networking and secure AI infrastructure. Cisco's commitment to incorporating Silicon One and SuperNICs into AI networking platforms, along with its Secure AI Factory, demonstrates a comprehensive approach to building end-to-end architectures that align with open principles.\n\nOverall, the expanded Open Secure AI Alliance at Black Hat emphasizes the importance of collective threat intelligence and open runtime guardrails in securing agentic AI. By promoting open standards and collaborative security practices, OSAA is paving the way for a more secure and efficient AI landscape, where enterprises can leverage open tools and frameworks to build robust defenses against AI-powered threats.",
  "summary": "The Open Secure AI Alliance introduced SAFE guidelines and open agent-security tools at Black Hat, giving enterprises a framework for safer AI deployment. The post Open Secure AI Alliance Expands at Black Hat: What You Should Know appeared first on TechRepublic .",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}