Using AI in professional work creates ethical obligations that go beyond legal compliance. The EU AI Act sets the regulatory floor, but operating ethically means building practices that respect clients, audiences, and the broader ecosystem—even when the law doesn't explicitly require it.

Here is the practical checklist.

1. Transparency

  • Disclose AI usage to clients in your service agreements and proposals
  • Label AI-generated content on your website and social media
  • Maintain a transparency page listing the AI tools you use and how
  • Never present AI-generated work as exclusively human-created when the client's expectations assume otherwise

2. Data Handling

  • Never input client data into AI systems without informed consent from the client
  • Understand the privacy policies of every AI tool you use (does the provider train on your inputs?)
  • Use paid/enterprise tiers for client-sensitive work (these typically include no-training guarantees)
  • Prefer local models for sensitive analysis where cloud processing creates unnecessary data exposure
  • Document your data handling practices so you can answer client questions clearly

3. Quality and Accuracy

  • Always verify AI-generated facts before publishing or delivering to clients
  • Never present AI speculation as established fact—if the AI generates a statistic, verify it independently
  • Apply human editorial judgment to every piece of AI-generated content
  • Be honest about limitations—if AI produced something you cannot verify, say so rather than presenting it with false confidence

4. Bias Awareness

  • Recognise that AI models carry biases from their training data
  • Review AI outputs for stereotypes, assumptions, and narrow perspectives before publishing
  • Diversify your sources—don't rely on a single AI model for important analysis
  • Apply your own professional judgment to challenge AI outputs that seem too neat, too generic, or too aligned with dominant perspectives

5. Responsible Automation

  • Maintain human oversight over automated systems, especially those affecting clients
  • Design kill switches into automated workflows—the ability to pause or override any automated action
  • Test automations thoroughly before deploying them on real client data
  • Monitor automated systems for drift, errors, or unexpected behaviour over time

6. Value Representation

  • Price your services based on value delivered, not time saved by AI—if AI helps you deliver better results faster, the client benefits from the quality, and you benefit from the efficiency
  • Don't charge for AI-generated work as if it were manual labour—this is deceptive
  • Reinvest efficiency gains into higher-quality outputs, deeper research, or additional deliverables rather than simply increasing margins

7. Environmental Consideration

  • Use smaller, more efficient models when they're sufficient for the task (Flash instead of Pro for routine work)
  • Avoid unnecessary regeneration—refine prompts rather than generating dozens of variations
  • Prefer local models when possible—they reduce the cumulative environmental cost of cloud AI inference

The Bottom Line

Ethics in AI consulting isn't about following rules. It's about building a practice that you would be comfortable defending publicly. If you would be embarrassed to explain any aspect of how you use AI, that's a signal to change the practice—not to hide it better.