GLS Legal Ops Webinar: An AI Update

and we will start in three two one welcome Legal Eagles today we're diving into the AI Abyss where algorithms wear suits and chatbots build by the bite buckle up it's going to be a wild ride this is the latest in the GLS series of master classes and today we're going to be looking at the interesting topic of AI now whilst that initial introduction was generated entirely by AI I promise that the remainder of this presentation is entirely organically farmed GLS wisdom um couple of quick housekeeping notes before we get going one you're all invited to introduce questions to the discussion however for the sake of the recording could I please ask you to keep your microphones on mute and submit your questions via the chat function we'll then address those questions at the end of the presentation also uh to your heads up you are being recorded and a copy of this recording will be published on the GLS website within 24 hours of completion of the presentation so with that let's get going firstly what are we going to be talking about today firstly we're going to be looking at what AI tools are available for in-house lawyers we're then going to be looking at the implementations of those tools and AI in general for in-house legal teams we're going to get relatively metaphysical and try and Define what exactly we're talking about when we say Ai and then once that is defined we'll look at some tips tricks and guidelines for how best to adopt a AI into your business and day-to-day operations in a safe but efficient way so first things that to catch the eye with now an early conception of this presentation was in fact just going to be an introduction to the AI based legal technologies that have been introduced in 2024 but there were in fact so many AI tools that have been introduced that that simply ceased to become viable honestly it's at the point that there are so many that specifically dedicated to in-house legal teams that the mere Act of animating this slide almost broke my PowerPoint presentation um and the tools that are currently being displayed are only the ones that I have personally tried this gives you a real sense of the explosion of AI tools in the marketplace today to summarize there's been a huge number of those tools some of them are free some of them cost very little and some of them are incredibly expensive if these tools are of interest to you and given that you're attending a webinar entirely dedicated to AI for in-house lawyers I assume that they are of interest to you we will be sharing these presentation slides after the presentation each tool is hyperlink in the slide so that you can go to it and try it if it is something that catches your eye in addition we have created an entire blog dedicated entirely to AI tools for in-house lawyers so I would strongly encourage you to browse through that blog and these tools at your own Leisure and experiment with the ones that you think would be of the greatest benefit to your department for the purposes of this presentation however we'll highlight that AI legal Tech that's being currently marketed to in-house teams can be Loosely grouped into four key categories and themes these are contract review and due diligence tools those are tools that do the thing that most looks like lawyering to the lay person for example legal sifter is a contract review tool that GLS sells through the legal operations center the key thing for this tool is that it assists lawyers to review contracts faster and to a higher standard than they would be able to do with without any assistance what these tools do not do certainly none that I have seen do they do not replace the lawyer rather these are tools that enable your legal team to do their jobs more efficiently in the same way that Excel as a spreadsheet enables your Auditors to do your accounting better faster cheaper they don't replace the need to have an accountant or an auditor now you may have notice that across the bottom of the screen are subtitles these are subtitles in Spanish of what I am saying during this presentation these are tools that are not directly related to the legal teams but are of use to help you get your day-to-day things done better faster cheaper or in this case let enable us to spread the word of GLS and legal operations to a wider audience than we may otherwise have been able to do the second category Falls in sorry and This falls into our second category of tools these are admin support tools so this translation service that we've been talking about or subtitling service that is now built into Microsoft Office and we haven't had to pay anything extra for it it just appeared one day right we'll come to that particular issue with a bit more detail throughout the course of this this presentation second ones that are more directly related to the day-to-day business of an in-house legal teams would be something like bright flag bright flag is another tool that is available on the GLS Operation Center what this is is a tool that helps you to review the invoices that you receive from your external legal councel the idea of the tool is that it automates and does very quickly and to a high degree of accuracy the reviewing of all of those time entries that make up your legal council's bill the net result is that the average invoice ends up being reduced by 10 to 15% on on an Undisputed basis that is the law firms themselves don't even dispute that the invoice should have been 10 to 15% cheaper because what the tool is doing is highlighting where Time Records have been double entered or where you've been charged for the wrong project etc etc etc this is a thing that typically took General Council hours to do every month and not only is it hugely time intensive it's an incredibly boring and frustrating part of the job so having an AI tool do it better than you're able to faster than you're able to is a real pleasure to add into your kind of Arsenal of tools that are available to help your department operate the third category is litigation tools now litigation Discovery typically involves processing a huge volume of documents and correspondence very quickly as you're building your case or Defending Your Case so this is an era where AI can be particularly useful because again AI is very good at processing huge volumes of information sensibly with making very few mistakes so also it's a very easy thing for AI developers to charge for because it's very clear what your traditional cost for doing that work with a man used to be compared to what it costs in terms of time doing it with the tool like it's a very easy one for one comparison and you can see that the AI does it much faster into a higher standard finally what we call the Trojan horse tools and services now these are in some ways a real blessing in the sense that these are freebie tools that suddenly just appear in your life in the same way that our webinars now have Spanish subtitles so wonderful like it's a a real point of value that's been given to us without any cost the other side of that issue is though that this is a real risk area when it comes to managing Ai and your protocols for AI procurement within your business so these typically apply in those procurement scenarios where annual renewals happen without review or happen automatically the classic example being your Microsoft 365 licensing fees just get renewed without anyone thinking and without anyone negotiating with Microsoft because frankly you're not going to get anywhere when you do negotiate with Microsoft but all of a sudden there's now an AI tool in your in your company's it ecosystem so how are you going to manage the risks that are inherent to any AI system okay the final point to mention on this slide is that many of these AI tools that we're talking about so legal sifter bright flag etc etc etc these are all available on the GLS legal operations center and for many of them it is possible to have a free trial of them to see if they would work within your ecosystem so if this is something that would be of interest to you do visit the legal operations center and do reach out to your GLS relationship manager and inquire about a trial for yourself okay so that was a bit of a swift Whistle Stop tour through a coup through some of the Cutting Edge tools on in the AI front let's now have a look at what these tools mean for in-house lawyers firstly and probably cannot be overemphasized it certainly isn't something that people are being shy about emphasizing is that AI represents an opportunity to improve the operational efficiencies within your legal department they're also a way of adding or adding value to your business and present investment opportunities for that business a tools themselves can be used to streamline routine tasks such as contract review legal research and document management which significantly free or reduces your time and Manpower costs and frees up capacity within the department a key aspect about the AI tools that have has quickly come to prominence is that AI appears to be particularly useful when it comes to processing large volumes of data or doing highly repetitive tasks I exactly the kind of tasks that human beings tend to get bored of doing and as a result of being bored or having their attention Wonder tend to make a lot of mistakes on so the AI Kind of Revolution does seem to suggest that there will be a really good way of reducing the human errors that appear in the day-to-day operations of an in-house legal department different slightly underappreciated aspect of AI is that in a multilingual region like APAC many professionals are operating in their second or third languages so being able to leverage natural language processing and machine learning has enabled the communication standards of the region to be significantly improved first drafts are being done by Ai and those drafts are often better than the standards being achieved by kind of a human trainee lawyer can produce themselves never mind a human traine lawyer who's trying to work in their second language and secondly you've got embedded proof reading and translation tools which further reduce the incidence of errors and typos Within your documentation this has really assisted crossborder deal completion and has reduced incidence of disputes that typically arise from miscommunication AI also provides a really good way of introducing Advanced analytics and predictive insights into the management of your in-house legal department these this means that you're able to anticipate and mitigate legal risk more effectively enhancing the overall risk management strategies of your business finally increased capacity for strategic work what we're talking about here is that the AI is effectively doing the timec consuming business as usual task that sucks up so much capacity of a legal department dayto day these are all important tasks they do have to be done we can't just ignore those tasks so AI provides a way of doing those tasks in in more efficiently and thereby freeing up your legal team capacity for higher value strategic activities that more directly contribute to the growth of your business basically what we're saying is that AI is a tool that can greatly assist legal departments achieve those that Hallowed Ground now I'm sure some of you have seen this graphic before but it is a good one to explain the [Music] point what we're trying to do is get the typical traditional capacity and workload of a legal department from this status where the vast majority of the time and capacity is dedicated to to grind Zone I relatively boring repetitive business as usual administrative type tasks over to this status where all of those grind Zone tasks are being done more efficiently there thereby freeing up the capacity of your department so that you can dedicate that capacity to kind of breaking new grounds and new jurisdiction dealing with major events like Black Swan events I doing the Strategic work that the business has actually employed you to do this transformation from Old Law to new law kind of uh capacity Arrangement is in essence exactly what all of the GLS products and services are designed to facilitate so AI fits very neatly into to that approach to Legal operations however that is AI when it is implemented correctly and that is a massive qualifier it it can be implemented in a way that achieves these wonderful outcomes but there are risks and if implementation has to be done correctly otherwise very material cost costs and inefficiencies can occur so with that in mind we want to move over from the light side of the opportunities presented by AI to the risks presented by AI now that tonal shift does risk me appearing to be the party poopy here but the role of the in-house lawyer must go beyond simply being excited by upside potential we must be aware of the risks inherent to AI well specific to AI tools and Ai implementations and it will be the in-house legal team that have a material role in mitigating and pro those risks and protecting the business from them some of the major risks to start thinking about are for example compliance risk with the suite of AI specific regulations that are currently being um released around the world there has in fact been an explosion of AI focused or AI adjacent legislation in almost every jurisdiction in the world this creates a new compliance obligation on the business therefore there is a risk of non-compliance moreover there has in fact been so much AI litigation that arguably there's more AI legislation now than there are AI tools to aim it at so some quick notes on the major bit of legislation that we need to be aware of the biggest most recent event in the AI legislation world is the European Union's artificial intelligence act the way that we would describe this act is that is a risk-based complex multifaceted piece of legislation that builds on existing legislation like gdpr what the European Union is doing with this act is trying to engender what has turned the Brussels effect in the same way that gdpr became the global standard for data privacy um legislation well gdbr got published and became the global standard for data privacy legislation all other jurisdictions around the world passing variations on data protection legislation but those variations are aligned with the requirements of gdpr so that those jurisdictions can continue to trade with the European Union I the world's greatest um trading block and a similar approach is hoping to be engendered in relation to to AI so the EU act whether it will or not remains to be seen but it is certainly positioning itself to be that go global standard that all other jurisdictions align with in the UK AI legislation has genuinely taken a different approach to what's set out in the EU AI act in the UK responsibly responsibility for AI legislation has generally been devolved to sector specific Regulators on the basis that those Regulators are more agile and have practical expertise applicable to the legislation now it is worth mentioning that a like a overall AI act has been tabled in the House of Commons in the UK and it does appear that a new government will be elected in the in the very near future so that current status may change in the near future but that is certainly the approach is currently been taken in the UK in the US it's a similar deal there's no overarching federal AI legislation rather responsibility for AI related top issues has been devolved to applicable federal agencies confusing that picture though is that almost every state within America all except for Wisconson for some reason but particularly California are passing their own versions of AI specific related legislation so there's a huge raft of AI legislation that's coming out of America and frankly such a wide variety of approaches being taken that that itself might cause a risk Vector I it might be very easy to effectively have Arbitrage between jurisdictions the final point on legislation is to look at it on an international context the point here is that all countries seem to be taking a very active role in addressing AI legislation so there's a lot of bilateral and multilateral collaboration occurring between nations it seems like everybody's learned the lesson from the the development of social media platforms and Tech B mon like Google and don't want to be caught sleeping as the next big thing occurs all jurisdictions seem to be actively trying to approach legislation on AI and find finding International treaties to reinforce their positions this includes um the Zan conference that happened about a month ago which is direct bilateral relationships between People's Republic of China and the African Union Nations also the soul Summit that happened six weeks ago in in Korea where Australia Canada China France Germany India Indonesia Israel Italy Japan Kenya Mexico Netherlands Nigeria New Zealand the Philippines Korea itself Rwanda Saudi Arabia Singapore Spain Switzerland turkey Ukraine United Arab Emirates the UK United States and the European Union all past um binding binding uh declarations so like this is a live topic that is seems to have really excited the world's legislators taking a step back we can say that the general Trend with all of this legislation that is being passed is that there seems to be a desire on behalf of all of those legisl all of those law um writers to promote Safety and Security responsible Innovation and development equity and unlawful discrimination or Pro protect against unlawful discrimination and protection of privacy and civil liberties this is being balanced by a very enthusiastic effort to ensure that their jurisdictions remain competitive and are promoting uh AI Innovation within within the jurisdiction in preparing for this presentation I read that no fewer than 15 jurisdictions from France to Australia to Rwanda have all said that they want to be the world center of AI Innovation so it gives a sense of like the tactics being taken when it comes to legislation okay so that's risks arising from AI specific regulation in addition to that there Remains the general risk of compliance with the existing regul ations for example gdpr the California privacy Rights Act the and the equality act and indeed in Singapore the Health Products act all of these are not ostensibly AI specific legislation but they are compliance obligations that might be at risk by the use of AI within your business so it's something to keep close attention to another key risk is intellectual property this is probably one of the areas where the greatest risk is the Divergence of approaches being taken across the world at this point it seems that most jurisdictions or many jurisdictions are taking very different approaches to intellectual property and the concepts of what defines intellectual property who can own it and what role does Concepts like fair use have in relation to intellectual property because of this lack of clarity of the law and how it's being applied there is a high risk inherent to this area this is a dynamic area that is undermining the ability of In-House legal teams to grow business within for for their companies and to protect their value it's something that has to be addressed very carefully speaking about being careful probably one of the most famous issues currently being presented by the mass adoption of AI is the risk of information leakage the specific risks here depend on the nature of the system that you're using or talking about and how it is being implemented in particular a key risk that seems to have appeared or seem to have materialized is the contractual relationship with the provider of that AI system a very good example recently being what happened at Samsung at Samsung's computer chip division three different occasions in 12- month period three different employees all working diligently in what they thought was the best interest of Samsung uploaded what turned out to be extremely confidential information to um open AI uh chat gbt to help them do that task specifically we're looking at um testing protols for computer chips now in this way incredibly valuable intellectual property suddenly got released to open AI but this is not something that Samsung as a corporation could resolve because Samsung had no contractual relationship with open AI it was just that their employees had accidentally uploaded training data to chat gbt and so what Samsung tried to do was that board members of Samsung literally telephoned board members of open Ai and try to find a resolution and they found a bit of a compromise but it the point remains there's a real risk of data being leaked to suppliers that your business has no ostensible relationship with this kind of goes to a greater point that a problem with AI at the moment is that it's very new technology that is not fully understood by users of that technology or even the developers of that technology so part of the risk factors that need to be addressed by in-house lawyers is the fact that mistakes are going to be um mistakes are almost inevitably going to happen so it's not um good enough to just say don't do act you need to be thinking about how to mitigate an issue that has occurred or how to respond to an issue that has occurred when you don't know what that issue is going to be ahead of time okay a few more quick risks and we'll caner through with AI systems implementation carries with it an integration risk like can that AI system integrate into your existing it ecosystem one question second question is if it does integrate into your existing systems does that have an impact on kind of volume based metrics and pricing that's inherent to your traditional systems so a good example would be do things like accessing to data or um trade trading numbers start getting influenced by your AI system um I guess the case to look at would be sap versus Diego that occurred in the UK high court in 2017 in that case whilst it's not directly an AI case a similar issue has occurred sap want paying for additional license fees when third party systems access the data generated by Sap's system integrating AI into your ecosystem runs that exact issue is your AI system starting to take data from another supplier system and is that affecting what your expected budgeting for those operations was going to be probably more of an issue for the it it Department by itself but certainly a risk factor that the legal department needs to be involved in mitigating finally we're talking about the system complexity we mentioned the sheer complexity of the system is causing problems to itself AI systems appear to be prone to huc Nations I presenting information as fact even though that information has just been effectively made up it's prone to privacy and confidentiality issues disinformation toxicity kind of amplifying embedded biases and it has obviously major issues for copyright the problem with AI systems is that they're so complex that is very difficult for legal teams at the moment to understand what's going on and therefore their ability to mitigate those risks is being challenged in this context the concept of explainability needs to start becoming a bigger part of your day-to-day operations when you're engaging with a AI provider or an AI enabled s uh service we need to start using explainability statements a lot more often to enable you to assess the risks inherent to what are very complex systems getting to the end of the risk these risks oops sorry let me reverse that getting to the end of some of these risks as in-house lawyers we do still have professional legal standards some of those standards are themselves being challenged by the mass adoption of AI in particular you have a professional obligation to supervise non-lawyers involved in your practice arguably and many academics do argue this that includes supervision of AI systems that you're using in your practice it is not good enough to just say that the AI system made a mistake as inhouse lawyers we are and will remain responsible for the failings of any AI system we use the flip side to that is that it's not okay just to say you know what I can't trust the AI system I'm just going to ignore it and not use it as part of my day-to-day operations as lawyers you do have a professional obligation to stay up toate with available Technologies particularly where that Tech technology can make your services better faster and cheaper so arguably we're all profession professionally obliged to become competent users of AI tools and systems that are available to us okay moving on to governance many of the risks I've identified above are risks that arise from specific implementations of specific AI tools on top of that we need to start thinking about AI governance as a whole what I mean by that is that the best companies are not just thinking about mitigating risks from specific applications rather they're thinking strategically about what does it mean for a business to be adopting many many forms of AI simultaneously like is that itself a strategic risk or more importantly how can that be approach strategically to maximize the profitability of the business and manage the risks inherent on so many different systems being adopted simultaneously um last point in terms of governance and I've put this here mainly because it doesn't fit very well anywhere else and mainly because it seems to me to be an issue that's being massively overlooked in the marketplace right now is that AI does have a massive environmental impact the AI Revolution doesn't come at it come for free for example adding generative AI to Google searches has increased the energy consumption of those searches by a factor of 10 to the extent that AI based Google Searchers last year alone equated sorry not last year in 2021 equated to the entire carbon footprint of the country of Ireland so there are real environmental impacts of the mass adoption of AI and if this is something that's important to your business and your your engagement with the world it see needs to be considered carefully okay what exactly is AI like what are we talking about here if we're going to start protecting our businesses from these risks inherent to AI we need to be able to include sensible contra actual provisions and adopt appropriate policies if we're going to do that we need to be able to Define what exactly AI is unfortunately there is not currently any universal definition of what artificial intelligence means in the absence of a kind of global standard we have come across some attempted definitions that have tried to approach this the question from a very simp approach in this case artificial intelligence means any system that uses or is based on artificial intelligence this was like a real example from a real contract that GLS ended up marking up now basically this is so simple that it doesn't work and it's kind of self-referential and is a terrible piece of drafting like this is this quote applies but I can see where the lawyer who drafted this was coming from basically without were taking the kind of the approach taken by the US Supreme Court in jacobes versus Ohio I you'll know hardcore pornography when you see it you this person is thinking what I'll just say is we all know what artificial intelligence is we'll know it when we see it however taking this approach is probably too vague to be useful and certainly it's open to challenge which itself is unreasonably expensive to resolve so maybe a better approach would be to copy a definition from some of that major legislation that we were talking about well there have been a lot of efforts by our steamed lawgivers to provide good definitions of what constitutes artificial intelligence and some of them have been fairly reasonable but as always there is an issue when law writers are trying to Grapple with cutting a technology that they are not experts in this example is from the National Security and investment Act of the UK from 2021 for me this definition highlights one of the the typical issues that's coming up in some of the legislation in this space it's falling into the Trap of blurring the line between a machine statistical processes and human perception and cognition for a working definition we need to avoid getting into metaphysical debates about what exactly defines intelligence or thinking um this is a rabbit hole that will go on forever and you'll get trapped in if you dip your toes into it so legislation like the National Security Act has a real problem that inherently implies that the system has some kind of human cognition or thinking what our recommendation to you today is that you can make things a lot easier for yourselves by remembering that the purpose of this definition is not to unveil some kind of philosophical Essence rather we're using this definition as a real tool to help us manage the the risk of AI implementation within our businesses so with that in mind the evolution of the European Union artificial intelligent act and its definition can be quite helpful the earlier version of this definition avoided that metaphysical question but it did have some issues of its own firstly it referred to a specific schedule of approaches this made the definition susceptible to becoming overtaken or made redundant by developments in the technology for example the intelligent machine approach of the second generation AI Systems Incorporated fixed programming rules and assumptions which is very different from the large language models that are invogue today so there is a real risk that your definition is not will not keep Pace with the the development of the technology also this definition foregrounded the issue of inputs and you'll see this in a lot of legislation around the world at the moment we're talking about human defined objectives and um inputs made by a person the problem with the is that it unnecessarily introduces instability to the definition as it's fairly unclear what a human defined objective is moreover it's questionable whether human inputs are necessarily required for many AI systems at all and thirdly a definition of a system is not typically need a definition of the person that's using that system so moving on to what became the definition of the final act that got passed by the European Parliament we can see that the references to Annex one techniques were removed there was also a lot of debate about that question of inputs and in the end the the references to inputs remained nonetheless we're now getting to approaching something that can be kind of useful for our purposes it's removed some of the ambiguity of the assist of those earlier drafts and there's a light at the end of the tunnel we can see potentially a way to resolution for me I particularly like the definition created by Chris Easton who's a partner at in the technology um Department within field Fisher his definition really tries to well firstly it borrows heavily from the UK and UK EU legislation but it avoids those metaphysical traps it avoids some of the ambiguities of of those definitions by Sid stepping the issue of influencing or inputs and I we don't even need to get into the debate about who input what or what the outcome was and finally it includes a future proofing IE other approaches designed to approximate cognitive value this definition is something that we could confidently Implement into our Clause Banks and templates today and be fairly confident that it will still work in 10 years time for me though I would respectfully suggest that his definition does still have a minor vulnerability for me I think that anything that relies on establishing the intention is opening a potentially very expensive can of worms as its inevitably cost of Fortune and legal fees to ever establish what another human being was thinking at a particular point in time so for GLS we would use Chris's approach to avoid references to inputs and implied perception and the ambiguity of that original EU act we' also for um future use this technique to kind of pro future proof the definition um for our future use but we would just simply sidestep the issue of intention entirely and use a definition that is solely Reliant upon objectively verifiable fact I is this A system that uses any of these things and generates any of these kind of outcomes so now that we have a definition of AI what other things can an in-house legal team be doing with that definition to facilitate implementations of AI into their business whilst mitigating the risks of those implementations one point to note is that in this section I'm assuming that you're a business that is not looking to purchase AI products You're Not Sorry is looking to purchase AI products but you're not a provider or developer of an AI system as you in that case would have Regulatory and compliance and explainability obligations that are quite distinct so like most risks in a commercial business primarily we're going to address those risks by the use of our definition in the updating of our procurement contracts when we're doing this however it's helpful to clarify two categories of procured artificial intelligence as these will Define the applicable risks and responses to those risks firstly we have ai as a service I your business is actually trying to procure an AI system and it is the system itself that is the key deliverable of your procurement the second scenario is when you're buying an AI enhanced service so this is this means that you're procuring a normal service or what traditionally would have been a normal service but now has a service provider that may in some ways utilize artificial intelligence to perform that service in the first scenario what we're looking at to address or what you'll need to address in your contracts are regulatory considerations data set controls IP protections acceptable use policies capability statements and cyber risk management of obligations on that final Point what we're talking about here is that suppliers will be looking to retain rights to suspend your access to the AI system whilst also imposing obligations on you as a customer to ensure that the systems are protected from integrated from your Integrated Systems this can become incredibly expensive so if this is something that's going to be an issue to do give us a call and we'll talk you through the issues here because the costs of protecting AI systems from Bad actors is unbelievable okay sorry at this point I'm running at PACE I'm a bit conscious of running out of time in this presentation when it comes to AI enhanced Services what you're needing to be looking at is first ly um what sorry first issue is that a lot of suppliers are trying to limit their warranties in their supply contracts to the equivalent of the deliverable will match the specification this does not work in an AI context not least of all because most lawyers are unable to or don't have the time to assess the adequacy of an AI system specification or the language that's being used to describe how that system will integrate into the supplier Services rather the approach you want to be taking in the scenario is insisting that supplier gives specific warrantees or commitments to outcomes not specifications now this has been best practice for service agreements for many years but it's particularly acute given the complexities of AI integrated deliverables you want it to be clear that the supplier Bears the risk of AI failing and there must be a sensible remedial mechanism in your contracts to compensate you as a customer in the event of an AI failure obviously data privacy is an issue but how exactly is an issue for one thing there's a real risk that the introduction of AI into a service provision changes the relationship between who which party is the controller and which is the processor recent legislation and guidance in the UK strongly suggests that the supplier as will become the controller as they are defining the algorithms of their AI systems which is very controlly Behavior so you need to be alert to data privacy issues whenever AI starts being involved in service provision moreover there's a real value to data ownership now that didn't traditionally exist is your supplier able to use your business's information to train their systems this is a concept that gets snuck in a lot into service contracts and if it's an issue to you you need to start thinking about do we restrict the supplier's use of that information entirely or do we insist that if the system is going to be improved by our information that there should be some kind of fee reduction or profit sharing mechanism in place to allow us to benefit from that improved AI system finally if you are going to remember one thing from today's presentation I would strongly recommend that you at least update your procurement templates to include an explicit obligation on your suppliers to be transparent and notify you if they intend to use AI at all this obligation should include an obligation on them to provide you with enough information that you can reasonably expect it to assess the security and appropriateness of the AI that they're intending to use no matter what your service and um Goods contracts currently say this is something that should be introduced to them to help to protect you okay we've got 30 seconds to run through the final to-do list of an in-house legal team when it comes to AI implementations firstly wello contracts are obviously a key shield for the in-house legal department when it comes to Defending Your Business from AI risks it's not the only defensive tool available to you you also have policies you should be looking to implement clear policies on AI use and procurement within your business the isaca which is an IT security um Association that's well respected around the world did a survey in 202 three and found that only 10% of companies currently have any kind of formal AI policy in place at all that's despite the fact that 40% of their employees reported that they were regularly using AI now the opposite issue is you can't just approach this this problem by creating a policy that says you as an employee of this company should not use AI a number of banks have done this but it's a terrible idea because frankly everybody within the business just ignores it so what you're doing is you're creating a well you're you're you're creating a problem for yourself and blindly ignoring what is quickly going to overcome uh which is quickly going to result in a tangible risk not least of all that you're effectively hampering your ability the ability of your employees to become skilled in a thing that will generate profit for the business as mentioned earlier the best companies have ai policies that extend to AI governance I they're thinking strategically about how many forms of AI can be quickly incorporated into their operations simultaneously whilst um serving an overall strategic Vision moreover if AI is a product that you are selling or is a product that is going to be involved in the services that you're selling you need to consider how to ensure that your services are now safe by Design and have appropriate explainability statements for your customers so that you can be transparent about your operations contractual protections are wonderful but they're not going to stop the harm that your business will experience and um from reputational loss and loss of potential business if the public thinks that what you're doing is the wrong thing or you lose face in the company in the Public Public side note that in Singapore you can freely access the model AI governance framework which comes with impementation guidelines and a compendium of um use cases so you really have no excuse to be falling for failing on this front quick easy wins for you to do is to update your Clause Bank to include appropriate AI Provisions these should these don't take long and if you need any support on this side you can ask GLS hey AI versus m&a due diligence is a massive topic that was going to be incorporated into this presentation but we're already over time so what we're going to do on this particular topic is defer it to a later presentation um but if in the interim you do have any you're considering investing into an AI provider or a business whose value is materially influenced by the AI that they're using please do reach out to GLS and we'll support you directly finally training that isaca server that I mentioned found that despite employees quickly moving forward with the use of AI as a technology only 6% of them had any kind of training for their staff at all that includes training for teams that are directly involved in AI this is extremely irresponsible if your business is not providing such training you as an in-house lawyer and legal department should at least consider implementing some kind of training for your legal team and encourage the business to roll similar training out across the entire ecosystem there's two aspects to this firstly your team needs to be conscious of the risks that are specific to AI so that they are better able to help mitigate those risks secondly AI use is not the same as a Google search how competent you are at prompting an AI system will directly result in the well directly relate to the quality of the output generated by that system this does take real practice and it's something that is not it's not intuitive you do need to get better at this this is a skill set to learn like how to swing a tennis a tennis record correctly and the world is quickly dividing into the skilled prompters and the cooks and you definitely do not want to let your legal team be left behind as the cooks so to wrap this up there are a lot of tools and opportunities there's a lot of tools available to in-house lawyers that provide opportunities to Greater improve the efficiency of your department these tools do however come with risks and those risks can be mitigated but this will require proactive conscientious thought from the in-house legal department itself um if you do need any assistance with these you can ask GLS to help you we provide a huge number of consultancy support services in this area including program design Clause bank and template upgrades Readiness evaluations and negotiation support so you're not alone on this journey feel free to reach out if you want some support Additionally you can as always go to the GLS legal operations center where there's a huge variety of free knowhow resources and tools for you to use to help improve and help how to proove your operations with that I would normally pass over to Q&A but frankly already over time so I'm just going to say thank you very much and if you do have any urgent questions please just email me and we'll follow up offline goodbye