The most important thing first
For errors of ChatGPT and other AI systems, the liability does not lie with the AI itself, but with the company that uses it. Therefore, it is essential to check every output before passing it on and to only enter sensitive data into approved systems. Depending on the damage, a business liability, professional indemnity, or cyber insurance may be relevant.
Who bears the responsibility for AI errors?
Artificial intelligence cannot be held liable itself because it has no legal personality. Those who use it in their company remain responsible for the application and the decisions that arise from it. Therefore, every AI output should be critically reviewed before it is shared with customers, employees, or business partners.
Which AI risks are particularly relevant for companies?
- Incorrect information: AI can generate convincing but content-wise false texts, for example in contracts, technical specifications, or disclosures.
- Data protection violations: problems arise when customer data, employee data, or confidential information ends up in external systems.
- Copyright risks: AI-generated texts, images, or code should also be checked before publication, especially in marketing.
Caution with sensitive content
Inputting confidential data into external AI systems is particularly tricky. Companies should clearly define which tools are permitted and which information must never be entered.
Which insurance helps with an AI-damage?
| Insurance | Typical use case | What matters |
|---|---|---|
| Business liability insurance | Personal or property damage | The specific damage case must be covered in the contract terms. |
| Professional indemnity insurance | Financial damage from faulty services | Particularly relevant when a false AI output leads to financial loss. |
| Cyber insurance | Data protection incidents, cyberattacks, or IT security events | Only the benefits actually provided for in the contract are insured. |
There is currently hardly any standard AI-risk insurance. Therefore, it is worthwhile to look at existing contracts and whether current coverage fits today's use of AI.
How do companies reduce their daily risk?
- Establish clear AI policies so everyone knows which tools are permitted.
- Check results before using them internally or externally.
- Observe data protection and only enter sensitive information into approved systems.
- Regularly review insurance coverage because new technologies create new risks.
From our consulting practice, it becomes clear: those who consider liability, cyber topics, and risk management together identify coverage gaps earlier. That is where a structured review begins, before an AI input leads to an expensive damage.
Frequently asked questions
Is business liability sufficient for AI errors like those of ChatGPT?
When does a company need professional indemnity more often?
May I input customer data into an external AI system?
Do I have to review AI-generated content before publication?
AI offers opportunities for companies, but also liability questions, data protection issues, and new organizational duties. For companies, viewing liability in the larger context of cyber and risk management provides better protection against unnecessary consequences. Now is the right time for companies to adapt their protection to the use of AI.
Möchten Sie Ihren Schutz bei KI-Risiken überprüfen?
MZO AG