Parties in holiday homes are the risk most owners already price in: noise, neighbours, damage, and a review that can sit for years. Airbnb’s answer is a machine-learning screen that runs before the booking is confirmed. It looks at hundreds of signals on an entire-home request and either blocks the reservation or redirects the guest toward a private room or hotel-style listing.
The company has described the system in its own newsroom posts for several years. The December 2025 update confirmed that the heightened version would run again for New Year’s Eve in the UK, the US (including Puerto Rico), Canada, France, Spain and Australia. A lighter form of reservation screening runs more widely year-round.
What the system actually looks at
Airbnb does not publish the full model. What it has said publicly is that the system scores an attempted entire-home booking against a large set of signals in real time. The factors it has named include:
- Length of stay. Short stays of one to three nights are scrutinised more closely around high-risk periods.
- Distance from the guest’s home location. Local bookings are treated as higher risk.
- How far in advance the booking is made. Last-minute requests score higher.
- Listing type. Entire homes are the focus; private rooms are lower risk.
- Guest account history. Reviews, account age and past behaviour are among the signals.
If the model flags the request as higher risk, two things can happen. The booking is blocked outright, or the guest is shown alternative options on the platform that are harder to turn into a large gathering.
Around New Year’s Eve the rules tighten further. Higher-risk one- to three-night entire-home bookings are restricted, and guests who are allowed to book must accept a mandatory anti-party attestation. That is a click-through confirmation that disruptive and unauthorised parties are not allowed and that breach can mean suspension or removal from the platform.
When and where it runs
The heightened version is switched on for peak risk periods, notably New Year’s Eve, in the markets listed above. Airbnb has run versions of it since 2020. The year-round reservation screening is broader but less aggressive. The system is aimed at entire-home bookings. It does not watch the inside of the property once the guest has arrived.
What the published numbers say
Airbnb attributes the figures to its own systems. In the December 2025 newsroom post it said that similar measures the previous New Year’s Eve had prevented more than 20,000 people in the US and more than 3,000 people in Canada from booking an entire home. An earlier UK-focused announcement (covering the same period of measures) put the UK figure at approximately 7,800 guests blocked or redirected, including around 2,900 in London and around 470 in Manchester.
On outcomes, Airbnb states that in the five years since the global ban on disruptive parties and the related enforcement measures, the global rate of party-related reports has fallen by more than 50 per cent. Earlier statements about the New Year’s Eve measures specifically claimed a drop of over 60 per cent in the rate of party reports in the countries where the defences were active.
These are Airbnb’s figures. They measure reports the company receives and the bookings it says it stopped. They do not show every gathering that still happens, nor do they measure damage or neighbour complaints that never reach the platform.
What the system does not do
It is a pre-booking filter. It does not monitor noise, occupancy or behaviour inside the property. Once a booking is confirmed, the host is still responsible for the house rules, the stated maximum number of guests, and any permitted tools such as a disclosed noise monitor. Airbnb’s own guidance for hosts continues to recommend clear rules, accurate guest limits, and prompt reporting if a problem starts.
The company also reminds users that third-party bookings and unaccompanied stays by anyone under 18 are not allowed. It asks parents and other adults not to book on behalf of someone else.
What UK hosts still need to do
The AI is one layer. It reduces the chance that a high-risk booking reaches the calendar. It does not replace the controls that sit with the owner.
- Write house rules that state a maximum guest number and ban parties. Make them visible before the guest books.
- Keep the listing’s guest limit accurate. An inflated number is an invitation.
- If you use a noise monitor, disclose it in the listing. Undisclosed indoor cameras remain banned.
- Keep a clear record of guest messages. If something goes wrong, the thread is the evidence.
- Treat the platform screen as a filter, not a guarantee. Direct bookings still need their own checks: identity, payment, and a simple house-rule acknowledgement.
The system is useful because it acts before money changes hands. It is limited because it cannot see what happens after the key is collected. Hosts who already run tight rules, accurate guest limits and disclosed monitoring are the ones who get the most from it. The rest of the protection still sits with the owner.
Key sources
Airbnb Newsroom, “Launching our anti-party technology for New Year’s Eve stays”, 20 December 2025.
Airbnb Newsroom (UK), “Anti-party defences for New Year’s Eve rolled out around the world”, 23 December 2024.
Airbnb Help, global ban on disruptive parties and third-party booking rules.
Contemporaneous UK coverage of the 2024/25 New Year’s Eve measures (including figures for London and Manchester).