Tenet Security

Tenet Security is an enterprise security platform purpose-built to protect autonomous AI agents from manipulation, hijacking, and data exfiltration. Founded by veterans of Israel's Unit 8200 and former Cisco GenAI security researchers, the company emerged from stealth in 2026 with a $6M raise and opened its platform to all customers. It targets the growing attack surface created by AI agents that operate with elevated privileges across enterprise infrastructure.

What It Is

Tenet is a runtime security platform for the "agentic layer" — the layer of autonomous AI agents, MCP servers, and tool integrations that increasingly act on behalf of enterprises. The platform addresses a threat category the company calls "agentjacking," where attackers manipulate or hijack AI agents to execute unauthorized code, exfiltrate data, or compromise other agents. Tenet deploys a lightweight Guardian Agent sensor that watches agent conversations and correlates agent-level actions to operating-system activity in real time, blocking unauthorized actions before they execute — without proxies, gateways, or latency penalties.

Core Capabilities: Discover, Detect, Protect, Manage, Validate

Tenet organizes its platform around five operational stages:

• Discover: Automatically maps every AI agent, MCP, and tool integration via lightweight traffic metadata, delivering a full risk assessment in under three hours, including shadow AI.
• Detect & Investigate: Reconstructs complete agent sessions — identity, MCP and skill usage, tool calls, and intent — to detect threats, misbehaviors, and agentjacking in real time.
• Protect: Stops destructive actions before execution, blocking agentjacking, data exfiltration, and agentic phishing. Aligned with AARM and OWASP Top-10 Agentic Risks, including agent-to-agent attack scenarios.
• Manage: Provides continuous risk monitoring across developer endpoints, codebases, and production cloud infrastructure without requiring agents to be relocated or rebuilt.
• Validate: Automatically red-teams agents in CI/CD against OWASP Top-10 for Agents and MITRE ATLAS, plus emerging exploits from Tenet's own Attack Lab and global Honeypot Network.

Agent-Side Simulation Architecture

A distinguishing technical feature is Tenet's Agent-Side Simulation (patent pending), which runs predictive models inside the agent's own runtime — three steps ahead of any action. Before an agent executes a move, on-device models simulate where that action leads; if the outcome is dangerous, Tenet prevents it before execution. The company states this operates without sandboxes or high-latency proxies, preserving team workflows while providing runtime defense built from first principles.

Coverage Scope

The platform covers coding agents (including Cursor), cloud agents (including AWS Bedrock), support agents, and customer-facing agents. It addresses threats including secrets leakage, hidden exfiltration, PII leakage, intent drift, rogue sessions, and agent-to-agent compromise. The FAQ confirms visibility extends to MCPs and Skills connected to agents, with granular forensics identifying the specific user, agent, and environment involved in every incident.

Update: Emergence from Stealth and $6M Raise

Tenet's homepage and blog announce that the company raised $6M and opened its platform to all customers as of mid-2026. The company is backed by early investors in SentinelOne and a strategic advisory board that includes former CISOs from Robinhood, BNY, and MIO Partners, as well as the OWASP AIVSS Lead. A Threat Labs webinar on agentjacking attack research was scheduled for July 9, 2026, signaling active research publication and community engagement. The founders — CEO Barak Sternberg and CTO Nevo Poran — previously co-founded Wild Pointer, which the About page states reached $2M+ ARR serving Fortune 500 companies.