92% of Germans want an AI kill switch. Brands should ship one

92 percent. That is how many people in Germany want every AI system to come with an emergency off switch. Most will read that as a safety debate for regulators and labs. I read it as a brief for every team about to put a chatbot, an agent or an automated campaign in front of customers.

What Bitkom measured

German digital industry association Bitkom published a representative phone survey on November 10: 1,006 people aged 16 and over, polled by Bitkom Research in calendar week 41.

• Kill switch: 92 percent want one for all AI systems, above 90 percent in every age group.
• Rules: 85 percent want stricter rules for especially capable AI, and 80 percent see AI in the wrong hands as a danger.
• Opportunity or risk: 56 percent see AI as more of an opportunity.
• AI that plans, decides and acts on its own: 47 percent feel confident about it, 49 percent worried. Among 16- to 29-year-olds, 58 percent are confident. Among those over 65, 62 percent are worried.

Bitkom president Ralf Wintergerst's point, paraphrased: the more independently AI acts, the more important it is to keep human control.

The kill switch is a brand feature, not a compliance checkbox

Here is what I find interesting: asking for an off switch is asking for trust. And trust is a brand's most valuable asset. That is exactly why, as CMO at burgerme, I deliberately kept AI-generated content out of the main brand's communication. Not out of fear of the technology, but because a breach of trust costs more than any hour saved.

In practice: your customer does not care how your agent works. They want to know they can stop it. A visible "talk to a human" button in the chat. An order the agent prepares but the customer confirms. An email that says plainly that a person signed off. That is the kill switch the 92 percent mean, just sized as a button.

The catch: your audience decides how much autonomy flies

For marketers, the age gap is the data point that matters. Younger people lean confident, people over 65 lean worried. Roll out one AI agent the same way to every customer and you optimize for one segment while alienating the other. An insurer, a bank or a travel brand with an older customer base needs more visible control than an app for students.

So my rule: human in the loop at the approval step is not just an internal process. It belongs up front, in the customer experience. Invisible control builds no trust.

What I would do now

• List every customer-facing AI use (chatbot, agent, automated emails, dynamic ads) and answer one question for each: where can the customer stop it or reach a human?
• Make the exit visible: a clear button or sentence, not buried in the fine print.
• Tier autonomy by audience: where customers skew older, start with suggestions the customer confirms instead of actions.

Bottom line92 percent want a kill switch. The brand that builds it in where customers can see it earns the trust the agent alone never will.