AI phishing and deepfake scams: how attackers target Slovenian companies with AI
Attackers use AI to write flawless Slovenian and clone executive voices. How to recognize AI phishing and deepfake scams and protect your company.
AI phishing consists of fraudulent messages created by attackers using artificial intelligence — free of grammatical errors, in flawless Slovenian, and tailored to the individual. This is joined by deepfake scams: a cloned voice of a director over the phone or even a fake video call demanding an urgent transfer.
TL;DR — Key highlights
- The old sign "bad grammar = scam" no longer applies.
- To clone a voice, an attacker only needs a few seconds of recording, e.g., from a video on LinkedIn.
- The most dangerous is Business Email Compromise (BEC), where an attacker impersonates a director or supplier.
- Protection: technical filters, SPF/DKIM/DMARC, MFA, and verification procedures via a second channel.
How AI is changing attacks
| Before | With AI |
|---|---|
| Generic messages with grammatical errors | Personalized messages in flawless Slovenian |
| Manual victim research | Automated data collection from LinkedIn and websites |
| Fake email address | Fake voice or video of a director |
| A few targets per day | Thousands of personalized messages per hour |
Scenario: the fake director
An accountant receives an email from the "director" who is traveling, followed by a short call with his voice: a supplier needs to be paid 18,000 EUR to a new account today. Everything sounds authentic. Such attacks have already caused millions in damages in Europe — and small and medium-sized businesses are increasingly targeted.
How to protect yourself
- Confirmation process: any change of bank account or urgent transfer must be confirmed via a second, known channel (callback to a known number).
- Agreed-upon keyword for urgent management requests.
- Email protection: correctly configured SPF, DKIM, and DMARC and advanced filters in Microsoft Defender for Office 365.
- MFA for all accounts, so a stolen password is not enough.
- Regular training with phishing simulations adapted to the AI era.
Frequently Asked Questions
- Can filters recognize AI phishing?
- Partially. Modern filters analyze the sender, links, and behavior, not just the text. Therefore, technology and procedures together are key.
- How do I recognize a deepfake call?
- Pay attention to urgency, requests for secrecy, and unusual channels. Hang up and call the person on a known number.
- Is this really a problem for small businesses?
- Yes. Small businesses often have fewer controls over transfers, making them an attractive target.
Related posts
Business Continuity Plan: What a BCP for SMEs Must Include
What do we do if nothing works tomorrow morning? A 10-page BCP is all you need.
IT SupportIT Infrastructure Documentation: What Needs to be Recorded and Why
If it isn't documented, it doesn't exist. The minimum documentation scope every company needs.
CybersecurityTwo-Factor Authentication: Implementation Without Employee Resistance
MFA stops the vast majority of stolen password attacks. Learn how to deploy it quickly and without complaints.
