Best AI Security Solutions Software - Page 3

Ackuity - Agent Execution Guardrails

Ackuity - Agent Execution Guardrails is AI security software that protects autonomous AI agents by intercepting threats at runtime. Operating within the execution loop, it prevents malicious actions before they damage systems, governing workflows and decisions beyond standard chatbot interactions. Policy-driven enforcement monitors actions with dynamic risk scores, halting operations when threats are detected. It prevents data exfiltration, unauthorized API calls, and cascading attacks, while intercepting multi-agent threats and enforcing granular access control. Immutable execution logs provide forensic documentation for investigations and compliance. The platform supports threat models and custom policies, addressing scenarios like prompt injection, jailbreaks, unauthorized data access, multi-agent collusion, malicious tool calls, and memory poisoning.

Act Security

Act removes excessive access across your cloud, eliminating the attack paths every breach relies on.

Actualyze AI

Aegara Trace

Aegara Trace is a comprehensive AI observability platform designed to provide organizations with real-time visibility into the actions of their AI systems. By monitoring every read, write, transformation, and API call made by AI tools, Aegara Trace ensures that businesses can oversee and manage their AI operations effectively. Key Features and Functionality: - Total Visibility: Offers real-time monitoring of all AI activities, including data access, processing, output creation, external requests, memory storage, and decision routing. - Sensitivity Awareness: Detects when AI systems interact with confidential or restricted data, enabling prompt risk identification without hindering operations. - Broad Compatibility: Supports native integration with multiple AI providers such as OpenAI, Azure OpenAI, Anthropic, Gemini, Bedrock, Cohere, Mistral, and over 30 others via LangChain, ensuring seamless observability across diverse AI tools. - Privacy-Centric Monitoring: Focuses on behavioral metadata rather than content, ensuring that sensitive data like prompts, responses, and file contents are neither read nor stored. Primary Value and Problem Solved: Aegara Trace addresses the critical need for transparency and control in AI operations. By providing a unified control plane to monitor AI behaviors, it empowers organizations to: - Enhance Security: Quickly identify and mitigate risks associated with AI interactions with sensitive data. - Ensure Compliance: Maintain adherence to regulatory standards by having a clear audit trail of AI activities. - Optimize Performance: Gain insights into AI system behaviors to improve efficiency and effectiveness. In essence, Aegara Trace offers a robust solution for organizations seeking to maintain oversight and governance over their AI ecosystems, ensuring responsible and secure AI deployment.

Agen.co

Agen.co is the identity-native agentic governance platform for enterprises adopting AI at scale. It lets security and IT teams become AI-native in a secure way, anchoring every AI agent to the identity behind it, governing every action at the moment it happens, and maintaining a complete audit trail traceable to a named, accountable human. The Problem Enterprises are deploying AI agents faster than they can govern them. Employees use Microsoft Copilot, Claude, ChatGPT, Cursor, and dozens of other tools to connect to corporate systems - often through unapproved integrations and reused API keys that security teams have never reviewed. In most organizations, the substantial majority of operating AI agents are not inventoried and not subject to any runtime governance. The access controls built for human users operate at quarterly cadence; AI agents act at millisecond cadence. Agen.co fixes that gap. How It Works Agen.co operates across three functions: Discover, Broker, and Govern. Discover scans identity providers (Okta, Microsoft Entra, Google Workspace), sanctioned AI platforms (Agentforce, Copilot, Bedrock, OpenAI), cloud environments (AWS, Azure, GCP), and endpoints to surface every AI agent operating in an organization - including shadow AI. All agents are unified into a single registry with a 0–100 Trust Score. Broker routes agent-to-system connections through Agen.co's MCP gateway - 150+ in-house connectors, a three-day SLA for new requests, and OpenAPI-to-MCP translation for any REST or GraphQL API. Enterprise credentials are vaulted and proxied; agents receive scoped, per-task tokens. Govern issues real-time authorization verdicts (ALLOW, DENY, STEP-UP, MASK, or APPROVAL-REQUIRED) in under 30 milliseconds. The policy engine combines static role-based rules with dynamic behavioral anomaly detection. Sensitive data is masked at the gateway using reference-token substitution across approximately 90 PII, PHI, financial, and secrets categories. High-risk actions route to human approvers via Slack, Teams, or email, with step-up MFA required. Every action is logged in a signed audit trail streamed to SIEM. For coding agents on developer machines (Cursor, Claude Code, GitHub Copilot), AgenShield provides on-device governance at the LLM-tool-call boundary - blocking destructive commands before they execute. Compliance and Deployment Certified SOC 2 Type II and ISO 27001. Six compliance packs: SOC 2, ISO 27001, HIPAA, GDPR, PCI DSS, NYDFS. Available as SaaS, hybrid (data plane in customer VPC), or fully self-hosted. Customers include Cisco, Rapyd, and Apono. About the Company Agen.co is built by Frontegg, founded in 2019 by Sagi Rodin (CEO) and Aviad Mizrachi (CTO). Frontegg's B2B Customer Identity and Access Management platform - the seven-year foundation Agen.co is built on - serves hundreds of enterprise SaaS companies globally. Frontegg has raised $70 million from Insight Partners, Pitango, and Stripes, with headquarters in San Francisco and R&D in Tel Aviv.

Agent Access Security Broker (AASB)

Unbound is the Agent Access Security Broker (AASB) built for AI coding agents. Discover, assess, and enforce policy across every agent in your org.

AgentGuard

In a single click, with 0 complicated setup Save costs (tokens) with Claude Code: - smartly reference cache for already read files - compress shell outputs - better replacements for grep, find etc Guardian Agent for coding agents: - Smart disallow for sensitive data reading by coding agents - Nudge the agent to use updated versions of libraries instead of relying on memory - Block the agent from using compromised libraries etc (protecting against attacks or agent mistakes)

Agentic AI Protection

Radware Agentic AI Protection discovers, monitors, and secures autonomous AI agents across an organization's ecosystem. It addresses critical visibility gaps and security risks introduced by agents that operate, adapt, and chain actions without direct human oversight. The solution provides real-time, behavioral-based detection against unique threats like prompt injection, goal hijacking, and memory poisoning. Additionally, it features continuous security posture management (AI-SPM) via a dynamic Risk Graph Map to score vulnerabilities and complex multi-agent risk paths. The platform integrates seamlessly with major enterprise AI services and custom-built agents while ensuring compliance with leading regulatory standards.

AgentID

AgentKey (Agent Access Management)

AgentLock

AgentLock is a security gateway that sits between AI agents and the services they act on. Every outbound call, whether an HTTP request, an MCP tool call, an SSH command or a browser action, is routed through a policy engine instead of going out directly. Low-risk reads can pass automatically, while anything that writes, spends or touches production waits for a human approval, delivered to the web dashboard or as a push notification on the mobile app and granted or denied in seconds. API keys and tokens never reach the agent. They are stored envelope-encrypted in AgentLock's vault and injected only at the moment of execution, so a prompt-injected or misbehaving agent cannot leak a credential it was never given, and secrets are redacted out of every stored request and out of the audit log. Every request, decision and result is written to an immutable audit trail with the payload, the policy that matched, who approved it and when: the evidence teams need for compliance reviews, incident analysis and simply knowing what their agents did overnight. AgentLock ships an MCP server, so Claude Code, Cursor and any MCP-capable agent connect with a few lines of configuration, and TypeScript SDKs are available for custom integrations.

AIMetrical

AIR Security

Airship AI

AI Security Gateway

AI Security Gateway (AISG) is an AI governance gateway that provides data loss prevention (DLP), enterprise authentication (SAML SSO), role-based access control (RBAC), SIEM integration, and Hybrid VPC deployment — all in a single OpenAI-compatible proxy layer. It sits between applications and LLM providers (OpenAI, Anthropic, Google Gemini, Meta Llama, Mistral, xAI Grok, Groq, and more), scanning every API request for sensitive data and security threats, enforcing policy before anything reaches the model. Core capabilities: 1. Data Loss Prevention — Auto-redacts 30+ PII entity types in text and images (OCR). Prompts are cleaned before they leave your infrastructure. 2. Prompt Injection Protection — Real-time detection and blocking of jailbreaks, DAN variants, instruction overrides, and system prompt extraction attempts. 3. Recursive Loop Protection — SHA-256 fingerprint-based detection kills autonomous agent retry loops (LangChain, CrewAI, AutoGen) at the gateway layer. 4. SAML SSO — Enterprise single sign-on with Okta, Azure AD, Google Workspace, and any SAML 2.0 identity provider. Auto-provisioning and enforced SSO supported. 5. Role-Based Access Control — 4-tier role hierarchy (Owner, Admin, Member, Viewer) with 17 granular permissions. Organization model with invitation-based onboarding. 6. SIEM Connectors — Native integration with Splunk HEC, Datadog Logs, and Microsoft Sentinel. Real-time security event streaming with configurable event filters. 7. Hybrid VPC Deployment — Compiled Go proxy runs inside your VPC. Prompts never leave your network. Cloud dashboard manages policies via metadata-only telemetry. 8. EU AI Act Compliance — Hash-chained, tamper-evident audit records satisfying Article 12 requirements. Records contain only SHA-256 fingerprints — never raw prompts. 9. Intelligent Cost Routing — Smart routing across 8+ providers and 600+ models automatically selects the optimal provider. Hard budget enforcement per project. 10. Webhook Notifications — HMAC-SHA256 signed real-time alerts for security events. Push to Slack, PagerDuty, or any endpoint. Integration requires 2 lines of code (OpenAI SDK compatible). Python SDK available (pip install aisg). Three deployment models: Open Source (Apache 2.0) self-hosted via Docker, Managed Cloud, or Hybrid VPC. 1M free credits on signup, no credit card required.