We increasingly hear that the work of an analyst, project manager or customer service specialist in a company is being taken over by artificial intelligence (AI). The savings are visible immediately; the risks appear later. If cybersecurity has not been put in order first, cutting headcount can cost more than it saves.

Whether Latvian companies are ready for these changes was discussed on 7 October at the experience-sharing day “AI × Cybersecurity”, organised by the green and smart technology cluster “Greentech Latvia” together with “Bite Latvija”. The event brought together company executives, technology professionals and industry experts, including from the Artificial Intelligence Centre, Latvia’s IT security incident response institution CERT.lv and the Latvian State Radio and Television Centre.

VersaIntel experts also shared their experience. In the part of the programme on companies’ experience with AI and digital security, we talked about what we have seen in practice and took part in an expert discussion on what to do after a cyberattack. In this article we summarise what a company leader needs to know before entrusting an employee’s duties to AI.

More confidence than preparedness

The event presented the latest survey by “Bite Latvija” and the research centre SKDS, in which 750 Latvian company executives took part between May and July. It shows that 44% of executives consider information and communication technology (ICT) solutions important for company growth. The security those solutions rely on is a more modest story.

  • Executives rate their own company’s cybersecurity at 5.91 on average on a 10-point scale
  • 18% of companies have already experienced a cyberattack that caused difficulties (14% last year); among large companies the figure is 39%
  • 33% of companies have introduced two-factor authentication
  • 15% regularly train employees on information technology (IT) security
  • 10% engage independent experts to assess or audit their IT infrastructure

At the same time, 34% of executives say that nothing worries them when they think about cyberattacks, and the share of such answers has grown over the year (29% last year). A third of executives are not concerned about attacks, even though only 15% of companies train their staff on security. This is the environment AI tools are now entering.

When the employee goes, part of the control goes too

An employee replaced by an AI tool did more than what was in the job description. The analyst noticed a figure that didn’t look plausible. The project manager asked why the account number on a supplier’s invoice had changed. That everyday vigilance was part of the company’s security, and it disappears along with the position unless someone has thought about what will replace it.

Along with the savings, AI also brings new risks into the company.

  • Data leaves the company. Employees enter contracts, customer data and financial calculations into AI tools, often from private accounts and in tools the company has not assessed. Management often does not know where this information is stored or who can access it.
  • Access rights are too broad. An AI assistant connected to email, documents and the customer database sees everything the user sees. If permissions have not been reviewed for years, the assistant will find in seconds what was never meant for everyone. In the survey, 62% of companies say they use Microsoft 365 daily, and AI assistants are now available in that environment as well.
  • Nobody checks the mistakes. AI answers confidently even when it is wrong. If nobody reviews the output, an inaccurate calculation or an invented fact ends up in a decision, an offer or a report to a client.
  • Attackers use AI too. Phishing emails are no longer given away by poor Latvian, and a manager’s writing style or voice can be imitated with freely available tools. 9% of companies have experienced phishing attacks on employees in the past year; among large companies, 32%.
  • The company becomes dependent on a supplier. The AI service provider becomes a critical supplier. If the service is unavailable or the account is compromised, a process that no employee knows how to do any more cannot be continued by hand.

What to put in order before handing work over to AI

Introducing AI is a management decision, not just an IT project. Before doing so, company leadership should be able to answer seven questions.

  1. Which AI tools are employees already using, including those nobody has approved?
  2. What information may and may not be entered into AI tools?
  3. What can the AI tool access, and have access rights been reviewed beforehand?
  4. Which employee is responsible for each process entrusted to AI, and what exactly do they check?
  5. Where does the AI service provider store data, and how will it act in the event of an incident?
  6. Have employees been trained to use AI safely and to recognise AI-generated fraud?
  7. How will the company operate if the AI service is unavailable for several days?

The last question is not theoretical. In the survey, 24% of executives admit that two to three days of computer system downtime would cause major losses for their company; among large companies, 75% think so. The more processes a company hands over to AI, the greater this dependency becomes.

If some of these questions have no answers yet, there is no need to give up on AI. It is worth starting the rollout by putting security in order, because doing it later costs more.

If an incident does happen

At the end of the event, experts discussed what to do after a cyberattack, responsibility, communication and business continuity. In the first hours of an incident, what matters most is whether it has been decided in advance who makes decisions, who stops the attack from spreading and who talks to clients and authorities. In a company where part of the work is done by AI, this plan must also cover how to cut off the AI tool’s access and continue the process without it.

How VersaIntel can help

VersaIntel helps companies put these questions in order before AI enters everyday processes. Our outsourced Chief Information Security Officer (CISO) assesses risks and sets security requirements, and in an audit we check access rights and system configuration, including in the Microsoft 365 environment. We assess AI service providers the same way as other suppliers that matter to the business, train and test employees with practical phishing simulations, and in the event of an incident help contain it, investigate it and restore operations.

Not every question requires a contract. If you are planning to entrust a process to AI, or have already done so, talk to our experts.

Request a consultation

The article uses data from the “Bite Latvija” and SKDS research centre survey “Use of ICT solutions in companies and exposure to cyber risks”. The survey was carried out from 19 May to 15 July 2026 among 750 Latvian company executives.