The IOC wrote its AI rules first. You should too
Sam Cherry
Founder
,
2403 Group
9
minutes to read
KEY TAKEAWAY
Responsible AI is what gives you licence to use AI with athletes, artists and fans, and it needn't slow you down. The IOC agreed its principles first, kept people in charge of the decisions that matter and published what it found. Any rights holder can copy that order of work on a much smaller budget.
AI usually arrives quietly: someone tries a free tool, it saves them an afternoon, and soon half the office relies on it without anyone setting a single rule. In April 2024 the International Olympic Committee did what most rights holders still haven't, and wrote down how it would use artificial intelligence before it used it at scale.
The Olympic AI Agenda set out where AI should help sport and the rules it would have to follow. Three months later the IOC put those rules to work at Paris 2024, using AI to protect athletes from online abuse. The technology was the easy part; the governance around it is what made it usable, and any rights holder can copy that.
2.4m
posts and comments screened by AI for abuse aimed at athletes and officials at Paris 2024
IOC via Deseret News, 2024
46%
of people globally are willing to trust AI systems
KPMG and University of Melbourne, 2025
8
governance principles underpin the Olympic AI Agenda
IOC, 2024
What's actually in the Olympic AI Agenda
The Agenda names five focus areas, running from supporting athletes, clean competition and safe sport to growing engagement with people and driving efficiency across the IOC and sport. Underneath sit eight governance principles: responsible, safe and secure, robust and reliable, accountable, accessible, fair and impartial, transparent and respect for privacy. Most of that list would sit happily in a club's own policy.
Two lines deserve a spot above every commercial director's desk. The first: 'Each user must be informed how their data is being collected and how AI decisions or outputs may affect them.' The second: 'AI must be free from latent bias and partiality.' Neither calls for a data science team, and both call for someone senior to own them.
How Paris 2024 kept people in charge
At Paris 2024 the IOC used AI to sift through some 2.4 million posts and comments aimed at the 10,400 athletes and officials at the Games, in more than 35 languages. The system flagged more than 152,000 for review, and people then verified more than 10,200 as abusive and reported them to the platforms.
The same model carried on to Milano Cortina 2026. Across the two Games the IOC identified nearly 25,000 abusive messages from 4.3 million posts and comments. AI flagged about 242,000 for human review, and around 24,700 were verified as abusive. The numbers are big, but the real lesson is the split: AI did the sifting, and people made the call.
If you can't explain the decision to the fan on the receiving end, don't hand it to a machine.
What this looks like at a club or festival
Most clubs, festivals and museums will never run Olympic-scale monitoring. They'll use AI to write copy, cut highlights, personalise offers and answer fans, and the same four questions apply. Consent: did the athlete, artist or fan agree to this use of their data? Likeness: who approves an AI-generated face, voice or style? Bias: who checks outputs for groups the model treats unfairly? And transparency: does the fan know when they're dealing with a machine?
These are fast becoming legal questions too. The EU AI Act entered into force on 1 August 2024, and the European Commission says its transparency rules take effect from August 2026, so people must be told when they're talking to a chatbot and deepfakes must be clearly labelled. The most serious breaches carry fines of up to EUR 35m or 7% of worldwide annual turnover. If you have European fans, partners or broadcasters, you should check where the Act applies to you.
Trust is the commercial reason to care. Only 46% of people globally are willing to trust AI systems, and 70% believe regulation is needed. So you're starting with a sceptical audience, and a fan who feels tricked by a chatbot or a fake voice may stop engaging with the brand that put it there, and with the sponsor beside it. And once fans stop engaging, you stop learning about them, so you lose the very insight that tells you which digital activations are worth building.
Worth pausing on
Could you list, within a day, every place your organisation uses AI on athlete, artist or fan data? And is there anyone who'd be willing to put their name to that list?
Three Key Steps
You don't need a big committee for this, just a short list, a clear owner and the habit of checking.
1
Write a one-page set of principles
Start with the IOC's eight principles and cut them down to what fits your organisation. Then add two lines of your own covering consent and likeness for athletes, artists and performers, and get the board to sign it. Keep it to a page if you can, because a short document people actually read will do far more than a long policy nobody opens.
2
Map every AI tool and its risk
List every tool that touches fan, athlete or visitor data, and include the free ones staff started using without asking. Rank each one by the harm it could do if it went wrong. Anything that decides a ban, a refund or access should get a human check, just as the IOC kept people verifying abuse reports.
3
Label it and keep a record
Label AI-generated content and chatbots clearly, since the EU AI Act now expects it for deepfakes and machine conversations. Alongside that, keep a simple log of the tool, the data source, the owner and the review date for each use. It's dull work, but it turns an awkward question from a sponsor or a regulator into a five-minute answer.
If you can't explain the decision to the fan on the receiving end, don't hand it to a machine.
Gulf rules give you a ready-made starting point
The Gulf moved early on AI rules. The UAE Charter for the Development and Use of Artificial Intelligence, issued on 10 June 2024, sets out 12 principles, among them algorithmic bias, data privacy, transparency and human oversight. In Saudi Arabia, SDAIA's AI Ethics Principles rest on seven principles, from fairness and privacy to transparency and accountability, and they apply to public, private and non-profit organisations alike.
I'd treat that as an opportunity as much as a duty. Government partners, sponsors and broadcasters will increasingly ask how AI is used on fans, athletes and artists, and a club, festival or museum that can show principles aligned to the national framework and to the IOC's approach has answered the question before anyone asks it. That organisation is also easier to partner with on AI-powered content and activations, because a sponsor can see how fan data is handled.
Smaller organisations stand to gain the most, because they rarely have a legal team to review every new tool. A one-page policy, a register of uses and a named owner give them the same discipline the IOC built for Paris at almost no cost, and those habits travel well to rights holders in Asia and Africa who work with partners in the Gulf.
Questions for your next leadership meeting
Take these into the room and answer them honestly, even where the answer is 'we don't know'.
Do we actually have a list of every AI tool people are using here?
Who signs off when an athlete's, artist's or performer's likeness is used in AI-generated content?
When AI shapes a decision about a fan, does a person check it before it sticks?
Would our fans know if they were talking to a machine?
What would we show a sponsor who asked about our AI principles?
From 2,500 AED per person
Timeline
1 day
If there's one thing to take from the IOC, it's the order it worked in: principles first, use cases second, people in charge of the decisions that matter and results shared at the end. Get that order right and fans will keep trusting you with their data, and that trust is the foundation of everything you know about them. Doing it responsibly doesn't have to slow the returns down, either. A Tier 1 European sports club in a programme our team led saw a 4.8% EBIT uplift within 12 months across 10 AI initiatives.
The basics are within reach for most rights holders in a matter of weeks. It starts with a team that's fluent enough in AI to use it well and to spot where it needs a rule, and that's what our Mastering AI Masterclass is built to do.
Keep Reading
Sponsorship
9
min read
Do you actually know what your biggest sponsorship is worth?
Activation
9
min read
Signing the sponsor is the easy part
Fans & Content
8
min read
Women's sport has the audience. Do you have the content?
Data & Insights
8
min read
Who are your fans, really? Your sponsors want to know
AI
8
min read
Your sellers spend most of their week not selling. AI can change that
Innovation
8
min read
Why sport innovation programmes stall after demo day
Sponsorship
9
min read
Your next sponsor might be in a category you haven't sold yet
Activation
8
min read
The fan games sponsors will actually pay for
Fans & Content
8
min read
Fans don't wait for match day. Neither should your content
Data & Insights
8
min read
Your ticketing system knows your fans better than you do
AI
8
min read
You don't need a big league's budget to publish like one
Innovation
8
min read
Show it to fans before you spend a season building it
Sponsorship
8
min read
Most sponsors decide whether to renew long before you ask them
Activation
8
min read
NFTs crashed. Some digital collectibles are still worth doing
Fans & Content
8
min read
What sport and culture can learn from music's superfans
Data & Insights
8
min read
Your match data has buyers. Are you giving it away?
AI
8
min read
The tech behind World Cup offside calls is coming to every level of the game
Innovation
9
min read
Where rights holders fit in the Gulf's sports tech boom
Sponsorship
8
min read
Who's actually sponsoring in the Gulf, and what they're buying
Activation
8
min read
Why your fan zone shouldn't close after the final
Fans & Content
8
min read
Arabic-speaking fans want more. Most rights holders are still translating
Data & Insights
8
min read
You have the data. Do you have anyone to read it?
Innovation
8
min read
Millions are visiting the Gulf's museums. Where are the sponsors?
Sponsorship
8
min read
Sport takes 71% of sponsorship money. Culture is missing a trick
