73% of MCP servers have vulnerabilities. Anaconda just bought the company that found them.

Enkrypt AI scanned 268,000 tools across 25,000 MCP servers.

143,000 vulnerabilities. 73% of servers affected.

Anaconda acquired Enkrypt on August 4 — their third deal in four months.

This is not a tools story. It is a CISO story.

Your AI agents are calling tools through MCP connections that nobody has audited. Enkrypt's data proves it: prompt injection, tool poisoning, data exfiltration pathways — all sitting in production right now.

The timing is not accidental. EU AI Act enforcement went live August 2. Enkrypt's compliance automation maps NIST and EU AI Act frameworks into enforceable guardrails. Anaconda just turned a regulatory deadline into a product feature.

Every enterprise deploying AI agents needs to answer one question today: how many MCP servers are your agents talking to, and has anyone scanned a single one?

If the answer is no, you are running on trust, not verification.

Audit your MCP connections this week. Demand red-teaming results from your AI vendors. The 73% number is not theoretical — it is the baseline.

SOURCE: https://www.anaconda.com/press/anaconda-acquires-enkrypt-ai
VERIFIED: Anaconda press release (August 4, 2026), BusinessWire filing, Enterprise DNA analysis
SIGNAL: Enterprise AI agent security is the new attack surface. 73% vulnerability rate across MCP servers means every agent deployment is a potential breach vector.