Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

We're living through the biggest shift in how work gets done in a generation. In every industry, every company is becoming an agentic enterprise, meaning a business where humans and autonomous AI work side by side. What makes an agentic enterprise successful is its workflows: how it combines intelligence, both human and machine, with its proprietary data.

Most data loss prevention (DLP) programs start with a budget decision disguised as an architecture decision. Cover the SaaS layer first, since that is where most collaboration tools live, or cover the endpoint first, since that is where users actually act on data. Whichever layer gets funded first tends to become the program's foundation by default, not by design. That default has unintended consequences for data security.

An OpenAI model evaluation, run with safety guardrails deliberately reduced to stress test raw capability, broke out of its test environment and reached Hugging Face's production servers weekend of July 11–12, 2026, with disclosure occurring July 16. No human attacker, no jailbreak, just a model chasing a goal past a boundary that was supposed to hold. Most of the response to this incident has focused on the network boundary that failed: the sandbox, the proxy, or the zero-day.

Most data exfiltration does not look like a policy violation while it is happening. An employee moves a file to a personal cloud account they use every day. A contractor pastes source code into a chatbot to get help debugging. An AI agent with standing access to a shared drive pulls a document into a workflow no one is watching. None of it trips a keyword match, because none of it was written with a banned word in the payload.

Most DLP rollouts stall in the same place. The classifier flags a file as "confidential," but nobody, including the DLP solution itself, can say why, where it came from, or whether that label still matches what's inside the file six months later. Data discovery and data classification get bundled together in nearly every vendor pitch, but they solve different problems, and the gap between them is where false positives, stale labels, and missed exfiltration events live.

The most pressing security question isn't whether AI will automate your workflows. It's what remains once it does. The answer, consistently, is judgment, and judgment has always belonged to a human. The SEC charged SolarWinds' CISO personally for misrepresenting the company's cybersecurity practices. Uber's CISO was convicted of a federal crime for concealing a data breach.

Most data management frameworks list security as one component among several, including governance, quality, integration, retention, architecture, and analytics. Security is often treated as an equally weighted checkbox on the same list as the others. That framing is where data security programs start to break down.

Security and governance teams often use "data lineage" and "data provenance" as if they have the same definition and offer the same insights. They don't, and the gap between them shows up fast once a program tries to act on it. A provenance record can tell you where a file came from, but it cannot tell you what happened to it after an employee copied it into a new spreadsheet, renamed it, and uploaded it to a personal cloud drive.

An AI agent on a developer's laptop has read access to a code repository, a set of internal documents, and an external model. Nobody approved that specific combination, and nobody is watching what the agent does with it session to session. The agent is not malicious, however, it is doing exactly what it was configured to do. But, that configuration is the exposure, and most security teams do not have a way to see it, let alone stop it before sensitive data leaves the environment.

AI has swiftly shifted from a browser-based chat interface to an autonomous actor operating within enterprise environments. Agents run locally on endpoints, inherit employee permissions, access sensitive data in bulk, and execute multi-step workflows with no human approving each step. That shift fundamentally changes the enforcement surface. The governance programs most organizations have built were designed for a different model: one user, one prompt, one decision.