Отчёт Thales Group: треть компаний уже используют ИИ-агентов, риски ИБ растут

Thales Group, a multinational corporation, surveyed more than 3,000 information security specialists from 20 countries in its report on the state of information security and found: 34% of organizations already use AI agents, while 73% plan to implement them within the next year. An analysis of the report was published in July 2026 by the InfoWatch expert analytical center.

How many companies use AI agents

According to Thales Group, 34% of organizations already use AI agents in their operations, while another 73% intend to implement them within the next year. The survey covered more than 3,000 information security specialists from 20 countries.

In parallel, the Business Information Security Association (BISA) surveyed representatives of Russian companies about information security risks when using AI in work processes — the Russian context is considered separately in the report.

• Thales Group survey: more than 3,000 respondents from 20 countries
• 34% of organizations already use AI agents
• 73% plan to implement agents within a year
• Information security budgets for AI tasks are growing by approximately 50% per year
• A separate BISA survey covered Russian companies

Why security is falling behind

Agentic applications have sharply increased the speed and volume of data processed, and it has become noticeably harder for companies to maintain the necessary level of protection, the Thales Group report notes. The number of organizations that allocate a separate budget for information security specifically for AI tasks is growing by approximately 50% per year.

This rise in spending is a direct consequence of the fact that old control models were not designed for autonomous systems. Data within organizations now moves faster and in greater volume, which means there are more points where a leak or error can occur.

What exactly is the new risk

The main risk highlighted by the Thales Group report is the transformation of an AI agent into an internal threat. An agent operates within the corporate perimeter, accesses internal systems and data, and acts autonomously, so weak control over it is more dangerous than many external attacks. Already a third of organizations (34%) use such agents — often faster than they establish rules for their safe use.

"As agentic AI is used more, ways of ensuring internal security need to be improved, otherwise AI will quickly become an internal threat," is the conclusion of analysts at the

InfoWatch expert analytical center following their review of the Thales Group report.

What this means

AI agents are moving from experiment to a standard business tool: a third of companies already use them, and most plan to implement them within a year. But the pace of adoption is outrunning the maturity of protection, and without strengthening internal controls, an agent risks becoming the most vulnerable point of the corporate perimeter.

Frequently asked questions

How many companies already use AI agents?

According to Thales Group, 34% of organizations already use AI agents, while 73% plan to implement them within a year. The survey covered more than 3,000 specialists from 20 countries.

Why do AI agents increase information security risks?

AI agents sharply increase the speed and volume of data processed and operate within the corporate perimeter. Because of this, information security budgets for AI tasks are growing by approximately 50% per year, and experts warn of the risk of an agent turning into an internal threat.