AI is getting out of control: more than 90% of companies recognize that they are not trained to use it

The topic was debated in the VeeamOn Tour 26 Buenos Aireswhich took place this Tuesday with more than 500 attendees in the conference rooms of the Sheraton hotel in the City. With more than 550 thousand clients in the world, Veeam is a company that offers a platform designed to make backups, recover data and ensure the continuity of a company’s systems after being the victim of a hacker attack or if it suffered a crash for internal reasons.

“88% of organizations already use or are piloting AI agents, but only 7% consider themselves truly prepared for it”says a study carried out by this company.

What slows down the advancement of AI in companies the most is data management. It happens that in many cases AI agents access and publish data that should not be revealed. “95% say that data-related challenges have already slowed their progress in artificial intelligence,” the same report maintains.

“It is known that the first question an employee asks their company’s AI is How much does the manager earn? And the striking thing is that in many cases it reveals the correct figure,” he told Clarion a Veeam executive in one of the conference hallways.

It happens that repeatedly no one controls what data a new AI agent that is being trained to, supposedly, enhance some company activity, has access to. They can access spreadsheets, personal emails of collaborators and executives, and medical files.

“90% of the data is unstructured, they are files, emails. Companies live like this. It’s part of everyday life. There is no control over an attachment that I send to you, whether that data is sensitive or not. And furthermore, you know that as a company you open the door of your organization to these agents. So you have to take care of what data you access,” he explained to Clarion Andrés de Beitia, Director, Inside Sales of Veeam.

The new control point is in the data: what information exists, where it resides, who can access it, what agents or applications use itwhat policies apply, and how to accurately recover from ransomware, human error, insider threats, operational failures, or unauthorized AI behavior.

In the region, companies are moving towards hybrid and multicloud environments, incorporating automation and artificial intelligence into their critical processes, while facing greater security, privacy and compliance demands. In this context, Veeam stressed that data resilience (ability to withstand and recover from interruptions or failures) can no longer be addressed in isolation: it must be integrated with the governance, security and trust necessary to operate in the age of AI.

With the certainty that AIs themselves can now filter data, Veeam bought more than $1.7 billion Securiti AIa Californian startup specialized in cybersecurity, privacy and artificial intelligence. With this addition to its portfolio, Veeam now offers a centralized data command and artificial intelligence platform (known as DataAI Command Center) to help businesses securely manage, protect, and comply with their data and AI models.

“The emergence of AI is equivalent to a tsunami. The speed at which it was deployed, the cycles are getting shorter… it’s all very fast. And the volume of information is now much greater. New data is generated, all interacting and generating data. So, as always, we have to take care of the data, but it is becoming more and more complicated because now we also have to take care of it internally,” closes De Beitia.