01
What AI can actually do today
Generated texts, voices, photos and video. We'll show examples of what can be made in minutes and why it changes work with confidence.
Course · Artificial Intelligence
Safe with artificial intelligence
AI writes, speaks and calls with your director's voice. We'll teach your team what to trust, what not to put in AI tools, and how to set up rules that stick.
For whom Management, authorities, companies Format Practical course Topics Deepfake, data leak, rules
Voice and video
a deepfake is created by an attacker in minutes
Data leak
what the team puts into the chatbots
Rules
short and really sustainable
Verification
second channel instead of intuition
AI security must be short, practical and sustainable
Practical impact
When voice and image can lie, a proven procedure must decide.
Artificial intelligence they are already changing jobs with confidence. The course converts deepfake, data leak and generated content into rules that the team can use on a regular day.
- 01
What doesn't belong in AI
Personal data, contracts, internal documents and school data must have clear rules, not just general recommendations.
- 02
What not to believe automatically
Voice, video or urgent email is not enough. The team learns to authenticate with a second channel and stop the fake authority.
- 03
How to put the rules into practice
Short procedure for management, teachers, office or company: allowed tools, prohibited inputs and output control.
What you will learn
02
Data leakage via AI tools
What employees put into chatbots and translators — personal data, contracts, internal documents. How to set what does not belong in AI.
03
Deepfakes and fraudulent scenarios
Fake director's voice, fake video conference, urgent payment. What these attacks look like and how to stop them is a process, not an intuition.
04
Rules for AI in the institution
Short internal rules that people actually follow: tools allowed, inputs prohibited, output validation.
05
AI and pupils
Generated content as a tool for cyberbullying and fraud in the school environment. What the school should be able to recognize and how to react.
06
Practical verification
Simple procedures to verify what you see and hear — back contact, second channel, follow-up questions.
Who will the course help the most?
01
School management and teachers
Generated content and fake accounts also target schools. We will teach the choir what not to believe and how to respond.
02
Authorities and public institutions
The official is the target of fraudulent calls and deepfake calls. We give it a simple validation rule.
03
Management and management assistants
The bogus instruction "from the director" is aimed right here. We train reverse validation before paying.
04
Financial and accounting roles
Urgent payment based on voice or video? We'll show you how to spot a scam and when to stop a transfer.
05
HR and internal communication
Sensitive data and communications attract both attackers and AI tools. We set what to never put out.
06
Social services and healthcare
Client and patient data is the goal. We will teach the team to protect them even under pressure and stress.
The principal's fake voice is not a script. It cost companies millions
This really happened. The attackers cloned the voice and image of management and forced employees to send money. This is exactly the type of fraud the course trains — how deepfake to know and how to stop it by process, not intuition. We do not guarantee that no one will try to scam you; we teach the team to verify before they click.
February 2024 · deepfake on video call
Arup — $25.6 million
An employee of the Hong Kong branch of engineering firm Arup joined a video call where the "CFO" and colleagues looked and spoke like real people. It was a deepfake. Believing he was following management's direction, he sent $25.6 million (HK$200 million) in 15 payments.
What we do with it: We teach the team not to trust the image or the voice — verify the request with a second channel and a call back.
Source: Fortune/CNN (2024)
March 2019 · cloned voice of the director
British energy firm — €220,000
The director of the British branch received a phone call from the "boss" of the German parent company — with the same accent and tone of voice. The attackers produced the voice using AI. Under the pressure of an urgent instruction, he sent 220,000 euros (approx. 243,000 USD) to an account in Hungary. The case was reported by The Wall Street Journal according to insurance company Euler Hermes.
What we do with it: Voice alone is not evidence. In the course, we practice verification using a known number and a control question.
Source: The Wall Street Journal / Euler Hermes (2019)
Early 2020 · cloned voice + fake emails
Bank transfer in Hong Kong — USD 35 million
A bank manager in Hong Kong received a call from a firm "director" whose voice he recognized, complete with emails from a supposed lawyer. He approved transfers worth $35 million for an acquisition that did not exist. According to court documents, the investigation in the UAE was described by Forbes.
What we do with it: A combination of voice and email bypasses intuition. That's why we build a process of verification, not trust in a single channel.
Source: Forbes (2021)up to USD 25.6 million
pulled off a single deepfake scam on a video call at Arup (2024) — the voice and image of management were fake
Source: Fortune/CNN (2024)Artificial intelligence up close
Deepfakes, supervision i training — where AI changes work with confidence.
What next
Let's build AI rules and training for your team
Send a request — who is to be trained, segment, date. We will come back with a customized program and calculation.
Safe with artificial intelligence in your institution
Send us a brief inquiry. We tailor the program for schools, offices and companies.
- Dlouhá 730/35, 110 00 Prague 1
- info@gbhdefence.com
- +420 252 548 480