Problem-free AI Success: How to Ensure GDPR-Compliance as a Freelancer

Artificial intelligence is on the rise and is revolutionizing the way we work. As an independent professional, AI offers massive opportunities to boost efficiency and focus on strategic tasks. However, using it carelessly introduces unforeseen and serious risks that you need to understand and manage.

Kevin Engelhardt, expert in cybersecurity and AI security, highlighted the five most important risks for freelancers and introduced a practical plan for the safe use of this technology in a recent Malt Academy webinar with our DACH team. The key lesson upfront: Harness the power of AI, but always stay in control!

It's vital to categorize AI correctly from the get-go. It’s neither a silver bullet, nor a threat that will make you obsolete: it’s a personal assistant. LLMs like ChatGPT are based on statistics and data; they are powerful tools that can support you in your daily work.

The most important thing to remember is to direct the AI and to always review its outputs carefully. According to the World Economic Forum, your unique human skills—analytical thinking, resilience, and creativity—will become ever more important in the coming years, as AI can only partly replace them. 

Careless use of AI can lead to serious consequences. You should be aware of these five risks and manage them proactively:

The most common risk is unintentionally uploading sensitive data into AI tools.

The problem: If you enter client or personal data, business secrets or source code into free or unsecured AI tools, this data can be used for model training. This exposes you and your clients to unnecessary liability risks, for example: by violating the GDPR.

Practical tip: Neutralise the data!
Before inputting confidential documents into an AI, abstract them. Replace real names, addresses, and company names with placeholders (e.g. “Person A” or “Project XYZ”). And most importantly: Don’t use free versions of AI tools. If there’s no cost, you and your data are the product.

AI models tend to convincingly present fabricated information as truth: this is commonly referred to as "hallucinations."

The problem: AI works with probabilities. An estimated 50–82% of all AI outputs contain at least one hallucination (depending on the study and model). An unchecked output that misrepresents legal or technical facts can greatly increase your reputation risk.

Practical tip: Fact-check & double-verification

• Check every output: Cross-check results with manual research and independent sources.
• Demand transparency: Ask the AI to provide sources and to rate its own confidence in its answers.
• Sparring partner approach: Compare ChatGPT answers with those from other LLMs like Claude or Mistral.

Because AI models are trained on huge datasets, often including copyrighted publications, there is a risk of plagiarism.

The problem: The legal situation is unclear and subject to change. When generating texts or images, it’s often unclear who owns the rights to the output and whether or not you’re allowed to use it commercially. This of course becomes even more complicated in an international setting across multiple borders and jurisdictions.

Practical tip: Transparency and AI ownership

• Use commercial licences: Choose business plans that contractually guarantee you explicit rights of use over the output.
• Mark content: Be transparent by labelling AI-generated material (especially images and videos) accordingly.
• Explicitly clean prompts: Ask the AI to create non-derivative content, i.e. not to reference any existing copyrighted works.

The convenience of AI brings about the human risk of somewhat blindly and mindlessly delegating tasks to it, without critically assessing which tasks are better done yourself.

The problem: If you only use AI as a “copywriter” instead of a “sparring partner”, you risk diminishing your own problem-solving capabilities and deep expertise over time. Much of the human contribution is lost if things are simply copied without thoroughly thinking them through.

Practical tip: Use AI as a brainstorming tool

• Start with your own ideas: Before using AI, take 10–15 minutes to think about the topic yourself and draft a structure.
• Focus on curation: Let AI create drafts or outlines. Always review, refine, and contextualize the output with your own expertise.
• Your expertise matters: Your experience and ability to think critically remain your most vital differentiator.

Lack of transparency about AI use can undermine your clients’ trust.

The problem: Clients may interpret the use of AI as a sign of lower quality, reduced value, or a lack of expertise. This is especially the case if AI is used “in the shadows”, without open communication.

Practical tip: Open communication and clarity

• Be transparent: Proactively communicate that you use modern AI tools to enhance efficiency.
• Use the phrase “AI-assisted, human-reviewed”: This shows that technology supports the work, but your human expertise guarantees final quality and approval.
• You are the final checkpoint: Make it clear that despite using AI, you are your own system of checks and balances, always reviewing and finalizing all results.

To help you integrate AI securely and responsibly into your workflow, Kevin Engelhardt developed the SAFER framework, giving you a clear structure:

S - Scrub Data
Remove all personal, confidential or sensitive data before entering your prompt.

A - Abstract
Use generic content and placeholders when making abstract requests.

F - Fact-Check
Check all statements and facts with a second tool or manual research.

E - Explain
Ask the AI to justify its answers and cite sources.

R - Rebuild
Review the final, quality-assured version yourself to avoid any liability and protect your own reputation.

The final message is clear: Invest in professional versions of the tools and maintain open, transparent communication with your clients. This way, you can make the most of AI as securely as possible and strengthen your profile as a future-oriented, independent professional.