Varonis: AI Data Ingestion Risks for Enterprise Security

Name: Varonis: AI Data Ingestion Risks for Enterprise Security
Uploaded: 2026-09-23T20:28:02-04:00
Duration: 52 s
Description: Discover how 3% of organizational data is being ingested into LLMs and AI, and how this creates a large data security risk.

Varonis

09/23/2026

0 (0%)

Report Like Favorite

TL;DR

Roughly 3% of enterprise data is already being ingested into AI models, creating significant and often underestimated security exposure for organizations.
AI adoption is forcing companies to recognize critical gaps in their existing data security and identity management strategies before deploying tools like Copilot.
The same capability that makes AI powerful — connecting disparate data points — is precisely what makes poor data governance so dangerous in an AI-enabled environment.

Summary

This short-form clip from Varonis highlights a striking data point: approximately 3% of enterprise data is currently being ingested into AI models and large language models (LLMs). While feeding more data into AI systems increases their intelligence and utility, it simultaneously expands the organization's attack surface and data security exposure. The speaker argues that AI adoption — particularly tools like Microsoft Copilot — has forced many businesses to confront long-standing gaps in their data security and identity strategies. When an AI copilot is granted broad access to organizational data, it can surface connections and insights that humans might never derive on their own, which is powerful but also dangerous if the underlying data governance is weak. The clip closes with a pointed question: is the data being fed into these systems even the right data? That question, the speaker suggests, opens an entirely separate and critical conversation about data quality, classification, and access control that organizations are only beginning to reckon with.

Chapters

0:00 - AI Data Ingestion Statistic
0:05 - More Data, More Risk
0:19 - Gaps in Data and Identity Strategy
0:45 - Is It Even the Right Data?

Key Quotes

0:00 "... 3% of the data that an enterprise owns is actually being ingested and fed into models of AI."
0:06 "The more data you feed into an AI, clearly the more intelligent it gets, but then that obviously then breeds more risk."
0:19 "A lot of companies have kind of realized just how either insufficient their current data security strategy is, or even their identity strategy when it comes to deploying something like a copilot."
0:38 "I can't think of another technology in our lifetime outside of cell phones, which had a slower adoption curve, that has been everywhere, everything all at once."

FAQ

Why does feeding data into AI models create a security risk?

The more data an AI model can access, the more intelligent and capable it becomes — but that same breadth of access increases the risk of sensitive data being exposed, misused, or surfaced inappropriately. If data governance and identity controls are weak, an AI tool can connect and reveal information that should have been restricted.

What should organizations do before deploying AI tools like Copilot?

According to the speaker, organizations need to evaluate whether their current data security and identity strategies are sufficient. The key question is whether the data being fed into AI systems is the right data — properly classified, governed, and access-controlled — before deployment begins.

Categories:

Tags:

Show more Show less

TL;DR

Summary

Chapters

Key Quotes

FAQ

Embracing AI Adoption While Ensuring Robust Security Measures

Risk in Real Time Demo Series: Virtual Patching: Protection at the Speed of Exploitation

Harnessing Data Governance for AI with Cyera and Snowflake