Skip to main content

Insights

What AI Can and Cannot Do in Close Protection

Open hand beneath a holographic AI chip, illustrating artificial intelligence in close protection

By Matthew Beer, Founder and CEO, King Safety and Security (KSS). Updated November 2026.

Artificial intelligence is changing how threats are researched and how attacks are prepared. It has not changed what keeps a Principal safe: sound judgement, careful preparation and the right people in the right place.

This article sets out where AI genuinely helps protective security, where it falls short, and the new risks it creates for families and the offices that look after them.

Where AI genuinely helps

Open-source research at scale. A thorough picture of a Principal’s public exposure can involve thousands of items: property records, company filings, press coverage, and posts by family members and staff. AI-assisted tools can sort and group that material far faster than a person working alone. Deciding what it means, and what to do about it, is still the analyst’s job. See Digital Risk.

Triage of monitoring alerts. Monitoring a location or an individual’s exposure produces a great deal of noise. Software can filter and rank it so that an analyst sees the developments most likely to matter first. See Intelligence Monitoring and Alerts.

Preparation. Summarising local reporting before a trip or drafting options for a complex schedule can save time, provided everything is checked against advance work and local knowledge.

At KSS, AI-assisted tools support open-source research and the triage of monitoring alerts. Their outputs are reviewed by analysts before they inform advice or protective arrangements.

Where it falls short

Context. Whether a person waiting outside a residence is a delivery driver, a photographer or a threat depends on context that rarely appears in data. Experienced operatives assess that context using observation, local knowledge and professional judgement.

Accountability. A recommendation that changes a family’s routine has to be explained and owned. A model cannot sit in front of a family office and justify its advice.

Presence. Advance work, relationships with venues and hotels, and the ability to act when a plan changes all depend on people on the ground.

Error. Automated systems produce false alarms and missed signals. Too many false alarms and people stop paying attention; too much trust in a quiet screen and real warning signs are missed.

Technology that identifies people raises further questions. Facial recognition can involve biometric data and significant privacy obligations, and any proposed use requires consideration of applicable law, necessity and proportionality (ICO).

How AI helps those targeting a family

The more pressing change is on the other side.

Research is faster. AI-assisted tools can reduce the time needed to assemble and analyse publicly available information about a family’s homes, routines and staff. The same tools that help analysts help them too. We describe how that picture is built in Digital Protective Intelligence.

Impersonation is easier. Voice and video can now be convincingly faked. In 2024, the engineering firm Arup lost around £20 million (HK$200 million) in Hong Kong to a fraud involving deepfake impersonation of senior colleagues (Computing). In a household, the equivalent is a call that sounds like the Principal, asking a house manager or assistant to change a pick-up, open a gate or move money.

Approaches are more convincing. Messages aimed at staff can now be well written, personal and plausible.

Practical steps for families and offices

  1. Agree simple verification procedures for any request to change travel, access or payments, especially one made by phone or video. Verify unexpected or sensitive requests through an independently established contact route, such as calling back on a previously verified number (Report Fraud).
  2. Brief household staff and personal assistants on impersonation, and make it clear they will never be criticised for checking.
  3. Review what can be learned about the family and its staff online, and reduce it where possible. A Personal Vulnerability Assessment is a good starting point.
  4. Treat anything an AI tool produces as a lead to be checked, not a conclusion.
  5. Ask security providers how they use AI, and who checks the output before it reaches you.

Frequently asked questions

Will AI replace close protection operatives?

No. AI can support research and monitoring, but protecting a person depends on judgement, presence and relationships that software cannot provide. Close protection remains a human discipline.

Does KSS use AI?

Yes, in a supporting role. AI-assisted tools help with open-source research and with triaging monitoring alerts, and analysts review their outputs before they inform advice or protective arrangements.

What AI-related risks should a family be aware of?

Impersonation is one significant AI-related risk for families and their staff: a convincing voice or video request that persuades someone to change plans, grant access or make a payment. Verification procedures are an important defence.

How can staff tell if a call is a deepfake?

Often they cannot, at least not reliably. It is better to rely on procedures, such as calling back on a known number or using an agreed code, than on spotting a fake.

Matthew Beer founded KSS in 2014 and has worked in protective security for more than twenty years, across UHNW households, family offices and corporate Principals in the UK, Canada and internationally.

Enquire Now

Contact us to find out how we can support you