Lasso Security vs Noma Security

Lasso Security provides an AI Security Platform that delivers visibility, control, and protection across AI models, agents, and applications for enterprises. The platform inventories AI assets including agents, models, prompts, tools, and guardrails via discovery and AI-BOM features. It offers AI Security Posture Management for misconfigurations, supply chain risks, and alignment with NIST and OWASP frameworks. Automated red teaming uses a library of over 3,000 attacks covering OWASP Top 10 and agentic threats like context poisoning. Runtime enforcement provides zero-latency decisions with remediation guidance, while AI Detection & Response leverages MITRE and OWASP for threat detection with 98.6% accuracy and sub-50ms latency, including intent analysis for anomalies. Lasso targets enterprises adopting GenAI, particularly regulated sectors, enabling secure innovation through real-time risk reduction, governance, and compliance support. The platform maps to frameworks like NIST AI RMF, EU AI Act, and ISO/IEC 42001 via runtime policies and audit trails, integrating with tools like Cloudflare and Palo Alto Networks.

Noma Security is a security and governance platform for AI agents and the wider AI lifecycle. It was founded in 2023 by Niv Braun (CEO) and Alon Tron (CTO), who met in the Israel Defense Forces' 8200 intelligence unit. Its contact page lists a single office, in New York, and its development center is in Tel Aviv. Noma has raised $132 million: a $7 million seed round led by Glilot Capital Partners and a $25 million Series A led by Ballistic Ventures, announced together when it came out of stealth (reported October 31, 2024), and a $100 million Series B led by Evolution Equity Partners in July 2025, with Ballistic and Glilot taking part. Databricks Ventures made an undisclosed investment in June 2025, alongside a partnership with Databricks. The platform has four products. AI-SPM finds agents, MCP servers, skills, models and tools across cloud platforms, SaaS platforms, code repositories and endpoints (through existing EDR or MDM tools), and flags misconfigurations such as excessive agency, unsandboxed agents and secrets in agent instructions. It also scans models from open registries such as Hugging Face for malicious artifacts, hidden instructions and known vulnerabilities, and assesses MCP servers for supply chain risk. Agent Access Control defines which agents are approved, what data they can reach and what actions they can take. AI-DR monitors agent sessions at runtime for prompt injection (including indirect injection through tool responses), data leakage, tool poisoning and scope violations; it can alert, block, mask data inline, steer the agent, require human approval or stop it with a kill switch. AI Red Teaming runs multi-turn attack campaigns against apps and agents, including in CI/CD pipelines, and feeds findings into runtime policies. Policies are enforced through agent hooks, AI and MCP gateways, agent SDKs and APIs. Noma covers SaaS agents built in Microsoft Copilot Studio and Salesforce Agentforce, and homegrown agents on Amazon Bedrock, Azure AI Foundry and Databricks or built with LangChain and CrewAI. Since February 2026 it has been available through the Extended plan of AWS Security Hub, covering Amazon Bedrock and SageMaker workloads. **AI compliance coverage** Noma says its AI-SPM and red-teaming findings map to the NIST AI RMF, the EU AI Act and ISO 42001, as well as the OWASP Top 10 for LLMs and MITRE ATLAS, and its red-teaming product offers these frameworks as compliance presets. A UiPath case study on its site says Noma's scanning supported UiPath's ISO/IEC 42001:2023 certification. We found no published control-by-control mappings. **Who it's for** Large enterprises with many AI agents across endpoints, SaaS and cloud that want one security layer from discovery to runtime. Noma says it is used by more than 10% of the Fortune 100, and testimonials on its site come from people at UiPath, Kantar, Best Buy, Nielsen and Endor Labs.