From utility to infrastructure: a shift in perspective
Only a decade ago, video conferencing was largely a support function. It helped reduce travel costs, enabled global collaboration, and served as a digital convenience. But over the past five years—and especially accelerated by global disruptions—video has become something else entirely.
Today, video is the boardroom. It is the site of m&a negotiations, public health briefings, strategic military decisions, and cross-border regulatory negotiations. It is the default environment for work that matters most.
Despite this evolution, organizational thinking about video remains trapped in the past. Procurement decisions are often driven by user experience and licensing models. Responsibility is pushed down to it and communications departments. The assumption is that video, like email or chat, is just a utility—easy to replace, inherently secure, not strategic.
This is no longer true.
The systems used to conduct video meetings are now part of an organization's critical communications infrastructure. They hold sensitive data. They carry legal risk. They are increasingly integrated with ai. And they must be governed accordingly.
Video is data—and AI turns it into more
The core misunderstanding about video is that it’s transient. We believe, wrongly, that what is said in a video call disappears once the call ends. But even before AI, this was never quite true—recordings were stored, logs were kept, and access data persisted.
With the introduction of AI, however, this relationship has fundamentally changed.
Every video meeting now creates multiple layers of data:
- The video and audio stream itself
- Metadata: participant names, timestamps, connection info
- Shared content: chat messages, slides, links, documents
- Ai-generated outputs: live transcripts, meeting summaries, action items, keyword tagging, and emotion analysis
These outputs are rarely deleted. They are stored in enterprise systems, cloud platforms, AI models—and often, without proper classification or governance.
The risk isn’t just that these materials exist. It’s that they are often stored in jurisdictions outside of your control, processed by third parties with different security postures, or even used to train models whose purpose is unclear.
And because these insights are helpful—they improve follow-up, create searchable archives, and power analytics—organizations rarely think about the tradeoff.
But that tradeoff is significant. Every derived insight from a video meeting is a copy of sensitive information. The more copies exist, the more exposure your organization has.
AI changes everything: four transformations
Artificial intelligence introduces capabilities that redefine what video meetings are and what they mean for organizational strategy.
1. AI changes how video is consumed
Real-time transcription, automatic translation, and live captioning make video more inclusive and efficient. These features enable global collaboration and accessibility. But they also mean that raw audio and video are being processed—sometimes on cloud infrastructure, using models you don’t control, with policies you didn’t configure.
2. AI creates new content
Meetings are no longer just recorded—they’re interpreted. AI copilots create structured summaries, pull out action items, rank priorities, and evaluate sentiment. These outputs are valuable. But they also introduce secondary data trails that must be governed like any other sensitive data asset.
3. AI enables synthetic risk
Technologies like deepfakes, voice synthesis, and real-time avatar mimicry mean that it is now possible to fake participation in a video call. Attackers can simulate faces, impersonate voices, and create entirely synthetic identities. Without strong identity verification, video trust collapses.
4. AI obscures the chain of custody
With AI-integrated tools operating across multiple layers—video, collaboration platforms, crms, and cloud storage—it becomes nearly impossible to trace where data flows. An AI-generated summary might be stored in one system, emailed from another, and cached in a third. This makes auditing and compliance extremely difficult—unless controls are built in from the start.
