Anaconda’s Enkrypt AI named Gartner sample vendor for multimodal protection
Enkrypt AI, now part of Anaconda, has been named a Sample Vendor for Multimodal AI Protection in Gartner’s Emerging Tech Impact Radar: AI Cybersecurity Ecosystem. The recognition covers security for AI systems that process images, text and audio, including systems that can use those inputs to trigger tool calls.
- Enkrypt AI is now part of Anaconda.
- Gartner named it a Sample Vendor for Multimodal AI Protection.
- The category covers AI security across image, text and audio inputs.
- Enkrypt AI says its tools test systems before release and protect them at runtime.
- The radar places Multimodal AI Protection in its three-to-six-year ring.
Why multimodal inputs create a security risk
Anaconda says AI agents can follow instructions hidden in images, documents, voice inputs and other data sources. In one example from the company, a customer service agent receives a support ticket with a screenshot containing an instruction to change an account’s payout details. If the agent follows it and calls the relevant tool, the account change could occur without an audit-ready log entry.
The company describes a similar risk in voice channels, where a spoofed caller could prompt an agent to approve a wire transfer. These attacks can involve image and text prompt smuggling, audio injection, transcription manipulation, cross-modal chains and poisoning of retrieved documents or stored memory.
Gartner’s definition, quoted by Anaconda, says that “Multimodal AI protection secures AI systems across all data types and modalities that they process.” The radar places the category in the three-to-six-year ring within its Securing AI, Runtime segment. Gartner says this range represents its estimate of the time needed for a technology to reach early majority adoption, defined as more than 15% target market adoption.
What Enkrypt AI says its platform does
According to Anaconda, Enkrypt AI tests multimodal systems before release and applies the same policy to text, image and audio inputs in production. Its Agent Red Teaming capability tests models and agents across 300+ attack categories, using adversarial image-text pairs, visual perturbations and audio distortions.
At runtime, Agent Guardrails apply policies across text, vision and audio. The company says its MCP Gateway enforces those checks inline on Model Context Protocol tool calls. Its Agent Policy Engine converts internal policy and regulation into a shared set of controls, while recording each guardrail decision with a policy ID, version and reason code.
In the support-ticket example, Anaconda says Agent Guardrails can approve or deny the account update before it runs. The company says the action is denied even if the model attempts it, preventing the hidden instruction from becoming an account change or wire transfer.
The recognition and its significance
Gartner’s Emerging Tech Impact Radar: AI Cybersecurity Ecosystem was published on October 5, 2026, according to Anaconda. The edition covers 22 emerging technologies and trends in the AI cybersecurity ecosystem. Gartner’s publication states that the radar assesses when a technology is expected to have a significant market impact and how large that impact could be.
Anaconda also points to Enkrypt AI’s earlier Gartner recognitions: a Cool Vendor in AI Security in September 2025 and a Specialist in the Emerging Market Quadrant for AI Application Security, Startup Vendors, in September 2026.
For users deploying AI agents, the announcement highlights a security requirement that extends beyond typed prompts. Anaconda advises teams to identify every place their systems accept uploaded files, images, voice, retrieved documents and MCP tool responses, then test those paths before production. Its stated aim is to help security teams find hidden instructions, stop unsafe actions at runtime and provide evidence of which policies were enforced.
