Built on the first Agent Gateway with native security, NeuralTrust protects what agents do, tool calls, reasoning, and MCP connections, not just what they’re told, without a custom integration for each agent or platform

LAS VEGAS, Aug. 5, 2026 /PRNewswire/ — NeuralTrust, the AI agent security company, today launched the first runtime security built to extend across every AI agent, without a custom integration for each one. Protecting agents has meant one of two things until now: a point solution built for a single tool, like a security add-on for one coding assistant, or a list of integrations rebuilt every time a team adopts a new agent or platform.

NeuralTrust inspects agentic traffic directly, through its own Agent Gateway, the first gateway on the market with native, built-in security, regardless of which agent or cloud produced that traffic.

Most AI security tools check what an agent is declared to do: a system prompt, a policy document, a list of permitted tools. NeuralTrust checks what an agent actually does, following its work through the full loop of requests and actions, not only the opening prompt:

The prompt reaching the agent, checked for injection attempts, exposed secrets, toxic content, and abnormal usage patterns.
Every tool call before it goes out, checked for code injection, such as a malicious database query, and exposed secrets.
Every tool or MCP response coming back, checked for indirect prompt injection and code injection hidden inside it, the point where an agent is most likely to act on something it never questions.
Every tool the agent tries to use, checked against authentication and authorization policies, so the agent can only access the tools and data it’s actually permitted to.
The agent’s own reasoning and tool selection, tracked as a loop across the full session, not one request at a time, to catch drift and runaway behavior.
The final response to the user, checked for policy violations, exposed secrets, and toxic content.

Alerts, audit logs, and analytics run across all six stages in real time and can be integrated with your company’s SIEM in one click.

“Every agent security product until now has been a point solution: built for one coding assistant, one framework, one vendor. That does not scale, because the agent ecosystem is not going to consolidate around a single provider. We built runtime security that plugs in once, through the gateway, and covers every agent that runs through it, without a custom integration for each one,” said Joan Vendrell, co-founder and CEO of NeuralTrust.

NeuralTrust has been named a Representative Vendor in two Gartner Market Guides, three Gartner Hyper Cycles and a Leader in KuppingerCole’s 2025 Leadership Compass for Generative AI Defense, recognition that reflects a wider shift: security teams are now expected to secure every AI agent in their environment, not only the ones a vendor has already built an integration for.

NeuralTrust is exhibiting at Black Hat USA 2026, booth 8106, through August 6. Technical documentation on TrustGate and TrustGuard, the gateway and detection engine behind this announcement, is available at docs.neuraltrust.ai.

About NeuralTrust

NeuralTrust is the leading platform for securing and scaling AI Agents. Recognized by Gartner, its platform helps enterprises identify, secure, and govern all the AI agents running in the company. Clients span from global banks and airlines to energy companies and government agencies. NeuralTrust’s mission is to make AI adoption measurable, governable, and safe, turning AI security into a strategic advantage.

Learn more at neuraltrust.ai

View original content to download multimedia:https://www.prnewswire.com/news-releases/neuraltrust-launches-the-first-runtime-security-mesh-extensible-to-every-ai-agent-302841146.html