Rise of AI Conference 2024 · Berlin

Responsible AI Use by Law Enforcement: Risks and Benefits

Irakli Beridze · UNICRI, United Nations

At Rise of AI 2024, Dr. Irakli Beridze, head of the Centre for Artificial Intelligence and Robotics at UNICRI, a United Nations agency, argues that governing AI is one of the major challenges of the current era, using law enforcement as the sharpest test case. He traces how, after organizing the UN's first AI meeting in 2015, when many delegations dismissed the topic as science fiction, AI governance became a UN priority, especially once Secretary-General António Guterres called AI a "game changer" in 2017. He describes UNICRI's work with INTERPOL since 2017 on a governance instrument for law enforcement, motivated partly by the sheer scale of data agencies must analyze, and highlights the AI for Safer Children program, which helps identify and rescue victims of sexual abuse. He then presents UNICRI's seven-part AI toolkit for responsible use of AI by law enforcement, developed over three years with EU funding, unveiled in Singapore in 2023, and tested in 15 countries, closing with next steps at the UN level, including a governance blueprint from the Secretary-General's Advisory Board.

Key takeaways

  • UNICRI organized the UN's first AI meeting in 2015, when many delegations dismissed the topic as science fiction.
  • UN Secretary-General António Guterres calling AI a "game changer" in 2017 helped legitimize AI governance work inside the UN, according to Beridze.
  • Roughly 45 zettabytes of data have been collected so far, projected to reach 175 zettabytes by 2025, more than agencies can interpret without AI.
  • UNICRI's AI for Safer Children program now serves 112 countries with more than 80 AI applications to help identify and rescue abuse victims.
  • UNICRI's seven-part AI toolkit for responsible law enforcement use was developed over three years, unveiled in Singapore in 2023, and tested in 15 countries.

Detailed notes

From science fiction to UN priority

Beridze recalls that UNICRI's AI and robotics work began in 2014, and that when he organized the UN's first AI meeting at the General Assembly level in 2015, many delegations considered it pure science fiction, with some formally complaining that he had brought a joke subject to the UN. He identifies three ingredients that make AI work: large amounts of data, computational power, and talented people building algorithms, arguing these only converged recently even though the notion of AI dates to the 1950s. He credits Secretary-General António Guterres, who called AI a "game changer" shortly after becoming Secretary-General in 2017, with validating his advocacy after two years of being dismissed internally. Since then, AI has become a UN-wide priority: the AI for Good Summit now runs in Geneva with 43 partner UN agencies, an inter-agency working group focuses on AI governance, the UN adopted a General Assembly resolution on safe, secure and trustworthy AI systems, and the Secretary-General created a High-Level Advisory Board tasked with producing an AI governance blueprint.

Why law enforcement is the hardest governance case

Beridze calls law enforcement one of the most sensitive and controversial applications of AI, balancing public fear over privacy and security with the reality that criminals also use AI tools. He notes the scale of data driving the need for AI: around 45 zettabytes collected so far, projected to reach 175 zettabytes by 2025, more than any law enforcement agency could interpret without AI support. UNICRI's Centre has worked with INTERPOL since 2017 on what it calls the AI for law enforcement movement, first getting countries to share information about the AI tools they use, then building understanding of responsible, human-rights-compliant use.

Real-world benefits: the AI for Safer Children program

Beridze highlights UNICRI's program combating child sexual exploitation and abuse, where AI classifiers help officers process seized devices containing potentially billions of abusive images and videos, a task that could otherwise take months or years and harm officers' wellbeing. He describes cases where AI analysis of hotel booking website images helped identify the specific rooms where abuse images were taken, leading to children being rescued. The AI for Safer Children program now services 112 countries and provides more than 80 AI applications.

Risks and the responsible-use toolkit

Beridze acknowledges that AI in law enforcement, if used improperly, can seriously harm human rights, erode the rule of law, and undermine public trust, which is essential for any crime-solving effort to succeed. In response, UNICRI spent three years developing an AI toolkit for the responsible use of AI by law enforcement, unveiled in Singapore in 2023 at INTERPOL's science conference together with INTERPOL's Secretary General. The toolkit consists of seven resources, was built through an inclusive process involving governments, law enforcement, the private sector, civil society and academia, and was tested in 15 countries over a year before finalization. Its principles center on minimizing harm, preserving human autonomy, and ensuring fairness. The European Commission funded the toolkit's first phase and will fund a second phase supporting countries in adopting the standards directly. Beridze notes the EU AI Act does not ban AI use by law enforcement but instead regulates it.

What comes next

Beridze says the UN's High-Level Advisory Board will present its AI governance blueprint by summer, to be introduced at the UN's Summit of the Future in September for member states to consider. He personally favors a treaty-based governance model with an implementing organization and verification mechanism, though he acknowledges this may be idealistic given how early AI governance still is. He closes by stressing that AI will keep developing rapidly and that the central policy challenge is ensuring the technology is properly governed so it delivers benefits without causing harm.

Quotes

AI governance is a major challenge of our times.

Irakli Beridze

Nobody should have any illusion that AI could be banned from the use of policing.

Irakli Beridze

Questions answered in this talk

What is UNICRI's AI toolkit for law enforcement?

According to Irakli Beridze at Rise of AI 2024, it is a seven-part governance instrument for the responsible use of AI by law enforcement, developed over three years with European Commission funding, unveiled in Singapore in 2023, and tested in 15 countries before finalization.

How is AI used to protect children from sexual abuse?

According to Beridze, AI classifiers help law enforcement officers process huge volumes of seized abusive images and videos, and AI analysis of hotel booking website images has helped identify rooms where abuse occurred, leading to rescued children; the AI for Safer Children program now serves 112 countries with more than 80 AI applications.

How much data does law enforcement need to analyze with AI?

According to Beridze, around 45 zettabytes of data had been collected by the time of the talk, projected to reach 175 zettabytes by 2025, an amount no law enforcement agency could interpret without AI tools.

Mentioned

AI governance, AI for Safer Children, responsible use of AI by law enforcement, EU AI Act, Summit of the Future

Keep exploring

More from Irakli Beridze

More from Rise of AI Conference 2024

More on Defense & Security

Transcript

Auto-generated transcript, not edited. · 3,917 words

Show the full transcript

thank you very much for this kind introduction and uh well hello everybody my name is Ira Le berit I'm as introduced head of the center for AI and Robotics for one of the UN agencies called unic uh this is probably one of the sort of oldest or longest running un initiatives when we started our work back in 2014 uh bringing the issues of AI at the United Nations level to discuss with me M states to ensure that basically the bottom line of our uh task or our work is to ensure that AI is used to solve problems but at the same time it is done in a responsible and human rights compliant manner so that concrete issues concrete problems are solved with use of powerful technology like AI but at the same time it's done in a proper way so that and it's a grandio challenge the name of my keynote the way I wanted to introduce was that AI governance is a major challenge of our times and I stand by these words because

I think that from this time to the foreseeable future one of the top challenges what we as a sort of uh humankind International Community need to solve is to solve the governance of this very very powerful technology why because uh the technology and AI as it's been demonstrated here by number of other speakers is growing exponentially and it will become more and more powerful and we do need to channel it for the sake of or for the sort of our surface so that we as a humankind will benefit from it and it has enormous potential as you all know it now I'm not going to sort of go into the definition of AI there is no formal definition of AI not by the United Nations or anyone else and there is no need for that it is a very powerful technology which probably mimics every behavior of the cognitive and others of the human beings but what is important is the three ingredients what makes Ai Ai and some other speakers already mentioned it it's

it's it's a lot of data it's a computational power and of course the talent the algorithm which uh very talented people are creating why it is important this three components was that while AI existed or or kind of notion of AI starting from 50s it has not fully taken off other than in some scientific circles especially not on the main main scale until we started to collect enormous amount of data manage the very powerful compute and uh and created algorithms with the support of the very very talented people and invested money in that now right now when basically when in in in 2015 I organized and I'll show you a slide about this a first meeting of AI at the UN General Assembly and this was not such a long time ago right I mean only something like8 nine years ago uh there were many un delegations who thought that this was a pure science fiction and there were some who actually formally complained about me that I brought a joke subject or a subject

which was a total science fiction to the United Nations and as the technology itself exponentially grew and uh now it's everybody talks about it the understanding and the need for the governance has also exponentially grow and right now we have the whole big race of how we manage the governance of the AI on all different levels here United States issuing a presidential act Chinese government is issuing different type of governance initiatives we had AI safety Summit at the UK with the 32 or so heads of States got together and started to talk about what to do with the with this powerful technology uh Summit AT Davos uh and I was present actually in both at the U AI Summit or at Davos uh Davos was fully dominated by AI I'm sure some of you went there this year and this was just all about AI almost and and I think and and I expected it will be the same thing in the next year to comes from the technical also from the policy or government

perspective as well what happens at the United Nations AI has increasingly became a top priority of the UN secretary General Antonio gutes when he became a Secretary General in 2017 and I can tell you a little anecdotal story about that one of his first speeches he mentions that AI is a GameChanger it can change our lives in a spectacular FAL and that kind of saved my career as well because I was advocating AI within the UN for two years and everybody thought I was crazy and then I said okay here is the secretary jail who also says that and then from that point on uh most of the UN agencies adopted AI programs AI uh leadership activities different uh uh task forces and at the UN we have numerous now ai initiatives we have ai for good Summit which is running in Geneva under the leadership of it with the 43 partner un agencies we have a inter agency specialized group on AI governance specifically and different organ different organizations are contributing to that and

at the end of course secretary General last year created an Advisory Board which is directly tasked to come up with the blueprint of the AI governance instrument totally it's a it's a very mam task it's very difficult task of course to bring uh uh not only The Advisory board members to agree but actually entire world I mean most of you are and we are at the sort of heart of the EU and you see how policy making is happening at the EU level or even National level and imagine how difficult it can be when you need 193 countries to agree on something and international policy making is a very difficult task in general and it takes us as an International Community long time to agree on certain things and especially on especially growing technology or changing technology like AI now this is the actually slide from that first meeting when we hosted the first meeting of the AI back in 2015 that's when a lot of people thought that this was this was an insane

initiative and why would un tackle it there were also questions that like you know why you bringing robots to the UN and a lot of people mixed it and they thought that AI was robots and uh and this was the moment when people thought that Terminator was Ai and that kind of and and and and and the interesting is that how much our understanding has been sort of refining from over and over and I think that I mean what ch GPT did and and that was the sort of an eye opener for many people said okay what is AI and for when you mention okay GPT there they said okay wow this is the real thing now uh on the governance side so un adopted a general uh UN General Assembly resolution on artificial intelligence safe secure and Trust forth CI systems encouraging everyone to uh pay a lot of focus to that that resolution why it is important is that the L and spirit of the resolution is to ensure that AI used in

a responsible and human rights compliant Manner and um and there is a general consensus about this how it is applied it's a different story and how this is going to continue it is also a different story but at the same but but as a least nobody says that we have to create AI which will harm other people and that's already a very good thing now on the side of the AI so what can be learned about governance through the lens of the AI for law enforcement and why we talk about the law enforcement law enforcement probably as you Pro would agree is one of the most sensitive application of artificial intelligence actually in any other fields right it's most controversial it's something that what uh general public is afraid of from both actually from the side of the privacy and and balancing with the security even there are aspects of how AI could be used by criminals terrorists and others and at the same time how to ensure that yes crime is needs to be

solved uh our Center is tasked to come up with both basically or actually three main tasks what we are dealing with and starting for many years now one is a governance instrument for the law enforcement to ensure that they use AI but in a human rights compliant manner it's a difficult task and at the same time to understand how criminals can use AI because there are numerous ways right now when criminals can actually conduct crime with support of AI tools and applications and at the same time to do I mean to do all of this in a systematic way so that we have all these tools available now starting from 201 17 uh we uh start working with Interpol and created something what we call AI for law enforcement uh movement and first task was to ensure that countries would cooperate and share share data and information what are the AI tools they are working on and how they solve concrete problems and the second was to ensure that they understand the AI governance from

the lens of the responsible use and human rights and they will have the available tools actually to do so so data I mean data is important and uh and we of course know this um data has we've been collecting enormous amount of data uh can I get a water actually from somebody so by now we collected uh around 45 zabt of data actually a little bit more than that and it is expected that by 2025 next year or year after we will correct 175 zabte of data which is a lot of data and uh no law enforcement officer no combination of the law enforcement agencies or anyone actually can properly interpret that amount of data that's why we need AI tools every single agency actually we spoken to in different parts of the world in different regions of the world either already using AI tools to solve concrete problems or are definitely interested in in that and I can even argue that every single crime or solution to the every single crime can benefit from

the a applications and there is no crime actually which anybody can tell me that cannot benefit from thei tools including a very simple crime areas as well now at the same time so therefore AI uh AI systems potential is definitely undeniable it boosts data and analyz completely repetitive task very fast it improves detection prevention and investigation cases and it safeguards the well-being of the officers as well our Center is running a uh very unique project on combating how AI can save children by combating sexual exploitation and abuse and uh through that uh and and I give you an example for example officers confiscate a computer or a device from an accused sex offender and then you have billions of images there horrible images children as young as few weeks old being abused and imagine that these officers need to go through all that data if not the AI classifiers which they using and they should be using it more extens Ely it takes people to actually go through image by image video by video and

it takes them month or even sometimes years to classify that data so not only that I mean they've been helped with these tools but at the same time uh their well-being is being saved as well now what are the real uh World benefits of um uh of the AI tools in the law enforc and then we have numerous use cases in that regard we have finding CH children from the sexual abuse materials and then over there uh our Center has sort of whole database of the use cases how uh this is working uncovering New Leads or even authenticating the evidence and uh and in that specific crime area the AI is probably the crucial part of any in investigative Solutions which is happening right now in the world uh there were cases when children been saved because AI application have identified through analyzing the data for example from the hotel uh booking websites and uh and some of the major booking websites cooperated with law enforcement agencies and provided the data and then run the

AI tools and actually found the specific rooms where the pictures were taken and children were saved through that so I certain does save children and uh our program which is called an AI for save for children which right now Services 112 countries in the world and provides uh 80 plus AI applications is a is a quite a powerful example of that now at the same time it comes with not without controversy there are a lot of of course data collected um uh not in a responsible Manner and then there is a lot of application which does threaten the uh sort of well-being of the of of all of us and obviously if not done properly introducing Ai and law enforcements enforcement does pose a serious concern and rightly so and there are many places in the world where this has been a case and this has actually caused a serious issues and it keeps sort of causing serious issues as well it can have a serious negative impact act on the human rights and uh

probably all of us understand how this can be done uh it can actually erose the rule of law and the and the principles and undermine public trust and I would focus a little bit more on the public trust side because for the law enforcement or or for anyone to to to have the uh real sort of a crime area solved with the support of AI you do need to have public trusting your law enforcement agencies doing so and that's sort of a ensuring that public trust has been always a very difficult uh place our Center is running another uh project which actually aims at uh supporting the sort of a bridging that gap between the public and the law enforcement to ensure that AI tools are properly understood and uh and also explained as well to the public that how the data is collected what is it done and and how uh certain sort of projects could be done that we would anonymize the data for example of anybody so that that people would not

be exposed to this um now how to reconcile the law enforcement needs for the Innovative AI tools with ai's potential risk and public distrust and we definitely need a proper governance we need governance tools we need those governance tools to be implemented in the countries and I'm talking about on the global scale and it is very necessary to have all of this done for 193 countries and it's a very difficult task so what happened was that uh when we started our uh sort of a joint work with Interpol uh and first task was to ensure that uh law enforcement agencies share the information about the AI tools they use and they have the data and support and the and the understanding how to use this properly but at the same time all of them were coming to us and asking that look we want to use the AI because we need to do it but at the same time we don't know what does it mean to use it responsibly and we don't know what

does it mean to use especially with a human rights compliant manner so then our task became to create a governance instrument for all of them and that's a quite a huge policy challenge uh so that uh on the one hand it would be supportive for their purposes at same time it would be human right compliant and it would be applicable to the 193 countries so last three years we spent actually creation creating this this governance instrument and what we call it a toolkit or AI toolkit for the responsible use of AI by law enforcement and we did unveil the toolkit in U 2003 last summer in June in Singapore at the Interpol science conference together with the Secretary General of the Interpol was there uh from from the UN side myself was there as well and uh I know there is a there will be a photo of the uran stock I think who was there as well uh during the unveiling and we very much appreciate the support of of of his as well

so what does this entail it is a it is a quite a comp complex and large document consisted of seven different uh resources and very practical and in principle if you take all of that uh and implement it in your National agency you should have a pretty good base and a cover to use AI in a responsible way but at the same time implementation portion becomes a very difficult task as well translating this um governance instrument into the national practices into the standard operational Pro procedures also becomes a quite a challenging actually undertaking and therefore our next phase of the initiative uh with actually we we are working very closely with European commission who funded the first phase of the initiative is going to fund the second phase of the initiative will entail working with regions with countries directly to help them to uh adopt the global standards on the responsible use of AI in the law enforcement there are countries who would be helping us and supporting us in doing so um by uh

and and those and there are countries who already supported in the creation of this governance instrument the process itself was very inclusive so we had stakeholders and different groups from all parts of the interested parties there we had government government representatives in separate subgroups we had law enforcement there we had private sector Civil Society Academia and many others and after we finished the uh creation of this instrument we tested it in 15 different countries for over a year so that they would use it for the national practices give feedback and then we would improve the improve the instrument so this is what I'm talking about how the entire sort of a toolkit approach or the governments was globally AC would would adopt the globally accepted principles politic principles laws ethics responsible AI Innovation and the main idea was that to ensure the good governance you need to minimize the harm you need to have a human autonomy and you need to have a fairness and many other initiatives which would be included in that as

well now the as I said it next stage of the next stage of of this sort of a global movement would be to ensure that this document will keep improving and at the same time will be implemented in as many countries as possible it could be sort of adjusted adopted but but at the same time it should serve a good base so that as I said it on the one hand police would need those tools and and again nobody should have any illusion that AI could be banned from the use of policing and we're very happy to see that EU AI Act also did not bend the uh use of AI by law enforcement it it will regulate it will ensure that it is used but at the same time uh we want to make sure that it is done in a in a proper and human right compliant manner now this is the U uh this is the available toolkit it has a website to it and all the governance instrument is is part

of it and uh this governance instrument became a one of the Practical examples at the United Nations level and at the global level how AI can be governed so we don't uh we and and certainly myself I don't say that this is the way or only way to do it there will be many different governance models on the AI in different fields and in in some Fields different things would work of course this is not a legally binding instrument so countries don't have to do it but at the same time this is a very good base to ensure that uh in a specific area especially in in in area which is so so sensitive and uh so in some cases controversial could be used as a model or as a governance model uh that countries can adopt and uh and uh and they would solve their issues uh there will be number of things happening actually at the UN level uh The Advisory Board which I mentioned will come up with the governance blueprint and

we don't know exactly what shape it will take whether it will be kind of a more treaty based mechanism which of the record I personally do prefer that type of governance models when countries agree they come adopt a treaty and they have an implementing organization or or even with the very very ification sort of a mechanism embodied in there so to ensure that everybody complies with that uh maybe it's very idealistic for the AI right now it could adapt more on the certain type of approach or it could adapt different type of approach but uh we will see it by the summer the shapes of of of this blueprint will be will be known and in September there will be Summit of the Futures at the UN and uh and the intention is that this model will be introduced there for the member states considerations and then next steps will be taken to there I think that we still have a long way to go before we will have a a serious governance or or

or sort of a very functional governance mechanism because uh you have the AI which which is developing and this morning people were talking about agis and whether it is possible or not and I'm not going to get to get into it especially not during this keynote but uh we will certainly see uh many developments on the technical side and AI is there only to grow and it will not stagnate obviously and uh and I think all of you agree with that and therefore our main challenge especially on the level of the policymaking is to ensure that this technology is properly go governed and to ensure that this technology actually provides the benefits and does not create harm for all of us with that I thank you very much thank you thank you

Rise of AI 2027 · Berlin · May 2027

Be in the room for the next one.

Ten years of talks started as conversations in one curated room in Berlin. The next edition brings 300 selected leaders together in May 2027.