Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

An account-takeover campaign against Instagram shows why agentic AI inherits every business logic blind spot we already had and then hands it a megaphone. Over the past weekend, a number of Instagram users, including the long-dormant Obama-era White House handle and a U.S. Space Force senior enlisted leader found their accounts hijacked. As reported by TechCrunch, the entry point wasn’t a stolen password, a phishing kit, or a zero-day in Instagram’s code.

Every company has a version of the same thing. Sometimes it’s a security wiki. Sometimes it’s a Confluence page. Sometimes it’s a PDF nobody wants to update.

Every week, someone in your organization stands up an AI service. Maybe they told security about it, but probably not. By the time it shows up in your inventory, it has been running for weeks, processing data, calling external APIs, and doing things nobody formally reviewed.

Traditional API security protects deterministic systems with known endpoints and explicit actions, while MCP-powered AI agents operate through inferred intent, dynamic tool chaining, and natural language interactions. This requires MCP-specific security controls such as tool governance, behavioral monitoring, and semantic anomaly detection.

This week, Connie Loizos, editor in chief of TechCrunch, sat down backstage with Francis de Souza, COO of Google Cloud, for a piece on the state of enterprise AI security. The interview is worth reading in full. Three points in it should reshape how every CISO is thinking about the next twelve months.

The enterprise security world is having two separate conversations that desperately need to collide. On one side, application security (AppSec) teams are scrambling to secure APIs – the connective tissue of every modern application. On the other, a new wave of “AI security” vendors promise to protect your LLMs from prompt injection, data leakage, and hallucinations. Both groups are solving real problems. Both are missing half the picture.

The non-human identity (NHI) problem was always the same problem: too many service accounts, too few owners, too many secrets in too many places. They sat where we left them, quietly piling up privilege, outliving the engineer who created them. Eventually someone, an auditor, sometimes an attacker, went looking and found them. Agents are a different problem.

Let’s start with a scenario. This is illustrative, not a single reported incident. A developer shares a Postman collection in Slack to move faster. “Here’s the Postman collection for the payment API. It has live auth headers so you can test prod endpoints.” The team uses it, work gets done, and the link stays. What no one realizes is that the collection lives inside a public Postman workspace. Weeks later, it is indexed by search engines. The URL requires no login.