New Security Platform Launches to Help Businesses Safely Use AI Agents
San Francisco, Monday, 13 July 2026.
Launched on July 13, 2026, JetStream’s new platform secures third-party AI tools, addressing critical trust concerns that currently prevent 83% of organizations from using AI at scale.
The Production Gap and the Rise of MCP
The divide between pilot programs and production-scale deployment remains a significant hurdle for enterprise artificial intelligence. According to UBS Evidence Lab data cited on July 12, 2026, a mere 17% of organizations use AI at production scale [1]. This leaves the remaining 83% of enterprises struggling to transition their systems out of experimental phases, primarily due to deep-seated concerns regarding the trust, governance, and auditability of autonomous AI agents [1]. At the center of this transition is the Model Context Protocol (MCP), an open standard defining how AI assistants securely connect to external data sources and enterprise tools [2]. Since its open-source release in November 2024, MCP has rapidly evolved into critical infrastructure for generative and agentic AI applications across major technology vendors [5].
Regulatory Pressures and Integration Hurdles
Despite massive developer enthusiasm, characterized by 97 million monthly SDK downloads, a massive gap persists as only 11% to 14% of AI pilots successfully reach production [5]. Security leaders face intense pressure to bridge this gap, particularly with upcoming regulatory deadlines. The European Union’s AI Act, which extends its reach directly to the tool-calling layer of AI systems, is set to become enforceable on August 2, 2026 [5]. This looming regulatory deadline forces compliance and security teams to seek robust frameworks that can govern AI interactions before non-compliance penalties take effect [5].
JetStream’s Solution to Secure the AI Supply Chain
To address these critical security and regulatory pain points, Santa Clara, California-based JetStream Security launched its JetStream Verified MCP catalog and JetStream AI Hub on July 13, 2026 [1]. The newly introduced platform establishes a comprehensive governance layer designed to manage third-party MCP servers and secure AI agent traffic [1]. AJ Anand, Co-Founder and Chief Technology Officer of JetStream, emphasized the urgency of the issue, stating that enterprises are no longer questioning the utility of AI agents, but are instead questioning whether they can trust those agents once integrated into vital corporate systems [1]. Anand noted that MCP raises the operational stakes because every connected server runs third-party code wired directly into core tools, data, and workflows [1].
Mitigating Vulnerabilities Through Attested Scans
The JetStream SAIG Platform operates as an OAuth-aware governance broker designed to manage MCP supply chains, discover unauthorized “shadow” MCP servers, and map approved servers to agents using JetStream AI Blueprints [1]. To eliminate supply chain vulnerabilities, the platform features more than 100 verified MCP images [1]. These images undergo rigorous semantic scanning to detect credential leakage, unicode smuggling, and obfuscated malware, with all scanning results stored as tamper-resistant, cryptographically attested records [1]. This enables administrators to apply granular runtime controls, allowing them to enable or disable specific MCP tool capabilities to strictly enforce the principle of least privilege [1].
Federal Standards and Security Frameworks
The deployment of these security solutions comes at a time of heightened scrutiny from federal cybersecurity agencies. The National Security Agency (NSA) recently released a Cybersecurity Information Sheet warning that the Model Context Protocol introduces substantial vulnerabilities, particularly regarding serialization, trust boundary hand-offs, and potential agent misuse [4]. To establish order amid fragmented proprietary architectures, the National Institute of Standards and Technology (NIST) officially launched its AI Agent Standards Initiative on February 17, 2026 [4]. Managed by the Center for AI Standards and Innovation (CAISI), this initiative aims to regulate autonomous systems across critical infrastructure by building industry-led standards and security research benchmarks [4].
Aligning Controls and Posture Management
Federal efforts to secure AI infrastructure have continued to expand throughout the year. Since March 2026, the National Cybersecurity Center of Excellence (NCCoE) has been actively developing frameworks to adapt identity and authorization protocols to the AI Risk Management Framework, specifically focusing on COSAiS SP 800-53 control overlays [4] [alert! ‘Federal adoption of these control overlays is anticipated but remains pending final implementation timelines’]. In tandem with these federal initiatives, security leaders are utilizing Q3 2026 industry symposiums, such as those hosted by IANS Research on July 15, 2026, to analyze MCP architecture, assess expanded attack surfaces, and explore how AI Security Posture Management (AI-SPM) tools can be deployed to safely govern enterprise resources [2].
A Growing Ecosystem of MCP Governance
The demand for robust AI governance has sparked rapid innovation across the cybersecurity sector. Just one day prior to JetStream’s announcement, on July 12, 2026, identity-access firm Datawiza launched its own enterprise AI governance layer, known as the Datawiza Agent Gateway [3]. Designed to provide precise MCP access control, the gateway acts as an authorization layer between AI agents and MCP servers [3]. It integrates directly with major identity providers (IdPs)—such as Entra ID, Okta, Ping, Google Identity, and AWS IAM—to validate signed tokens and enforce policies based on user groups, scopes, and custom attributes, all without requiring organizations to modify their existing MCP servers [3].
The Path to Production
As organizations navigate this rapidly shifting security landscape, the industry is preparing to showcase these new technologies at major global events. JetStream has announced plans to demonstrate its Verified MCP catalog and AI Hub at Black Hat USA in Las Vegas at Booth 4705, scheduled for August 3–5, 2026 [1] [alert! ‘The demonstration is pending the scheduled occurrence of the event’]. With the EU AI Act enforcement date approaching on August 2, 2026 [5], and federal standards solidifying [4], the launch of unified governance platforms marks a pivotal step toward moving autonomous AI agents out of the pilot phase and safely into production scale [1][5].