Many people assume that experience with keyword searches on Google will equip them to interact with an AI. Not so.

“Questioning is a key cognitive ability,” says Job Voorhoeve. He is joined by Jamal Khan, Managing Partner of Amrop in Australia: “AI is only as good as the instructions that you give it.” As Amrop Global Programs Director, Costa Tzavaras has been instrumental in integrating AI across the partnership. He explains: “Learning prompt engineering is important, as is understanding as much as we can about the black box and its responses. We have a professional responsibility. We are rolling out AI literacy for everybody. Prompt engineering. Ethics. What are the risks? The benefits?” He compares this to the advent of sustainability. “We had to ask: do we know what we're talking about? We have new concepts - carbon capture, ESG. AI is the same: literacy, then usage and skill development.”  

One risk is over-reliance on GenAI in leadership hiring: “You think you have the answer in front of you, instead of thinking yourself, and using different tools,” warns Amrop Global Board Member Mikael Norr. 

Confidentiality is the foundation of executive search, whether for client or candidate personal data, CVs, conversations, or processes. Even enterprise-grade AI requires human vigilance: security-first thinking, strict protocols, and responsible use. And as AI scales up, so will the ethical and regulatory concerns. 

At Amrop, every AI discussion begins with data security. It is embedded in internal protocols and strict enterprise licenses. “Our systems are not accessible by robots,” says Job Voorhoeve. The firm’s longstanding global GDPR policy includes the right to be forgotten. Data is ringfenced by confidentiality agreements and opt-outs. “A candidate has to understand that they have a choice, says Costa Tzavaras. “Amrop’s upfront consent process must include AI tools.”  

If AI tools can help researchers find potential candidates using public data (for example, on LinkedIn), Amrop never feeds personal or confidential information into external, open systems such as ChatGPT. The front and back door remain firmly bolted.  

But business leaders must be aware that the AIs who they invite to join confidential gatherings may be somewhat indiscrete. “I've been in meetings with boards concerning highly confidential recruitments where a board member has their own AI assistant recording,” says Mikael Norr. The output is automatically dispatched to all participants. “It’s suddenly in ten people's email inboxes, which is much more vulnerable than a meeting within four walls.”