Latest / Herbert Smith Freehills Kramer Podcasts / Legal Tech Deciphered EP7: Beyond Messaging Apps: The Rising Challenge of Social Media Data in Litigation
Transcript
- 0:00Roisin McCrory: Social media data is high value, but it's high risk and low certainty.
- 0:04Roisin McCrory: So the challenge isn't just access, it's around that authenticity and it's around trust.
- 0:09Roisin McCrory: Welcome back to Legal Tech Deciphered Trust, Truth and Hallucinations.
- 0:23Roisin McCrory: In our last episode, we explored one of the biggest barriers to using analytics and Gen AI effectively in disputes, and that is simply getting hold of the data in the first place.
- 0:33Roisin McCrory: Today we're taking that conversation a step further and looking at a data source that is becoming increasingly important in litigation, but also increasingly difficult to deal with, and that's social media.
- 0:44Roisin McCrory: Unlike traditional corporate data, social media content often sits outside organizations, it's on personal devices or third-party platforms.
- 0:54Roisin McCrory: It can be visual, informal, and difficult to collect, and sometimes it's even designed to disappear altogether.
- 1:01Roisin McCrory: In this episode, we'll explore why social media is playing a growing role in disputes and investigations, what makes it so challenging to preserve and collect, and how legal teams are adapting as platforms, technologies, and user behaviors continue to evolve.
- 1:17Roisin McCrory: I'm Roshan McCrory, Head of Disputes for the Digital Legal Delivery Team.
- 1:21Roisin McCrory: And joining me today are David Shepard, Senior Associate in the Disputes team.
- 1:26David Shepherd: Hi Roshin, great to be here.
- 1:28Roisin McCrory: Uchi Jay Ordenia, head of data analytics.
- 1:31Uchi Jayawardhena: Thanks, Roshin.
- 1:32Uchi Jayawardhena: Looking forward to this discussion.
- 1:34Roisin McCrory: And Megan Ryan, a senior manager in our eDiscovery team.
- 1:38Roisin McCrory: Hello all.
- 1:39Roisin McCrory: So between them, they will help us unpack the legal, technical, and operational challenges that social media presents.
- 1:47Roisin McCrory: David, why are we seeing social media data appearing more and more frequently in litigation today than we were, say, even a few years ago?
- 1:58David Shepherd: I think from a litigation perspective, there are probably two key drivers.
- 2:02David Shepherd: The first is that litigation is inherently backwards looking.
- 2:06David Shepherd: And so cases are often looking back at a period a few years in the past, and so a period uh now where social media usage has grown quite significantly, in particular for the sort of uh commercial disputes that we work on.
- 2:19David Shepherd: And then secondly, the courts usually place considerable weight on contemporaneous evidence, and social media can capture that in a really very candid way, in a really useful way, and so it can be extremely valuable evidence.
- 2:31Uchi Jayawardhena: And that's very interesting because the majority of users treat social media as informal.
- 2:37Uchi Jayawardhena: However, depending on the case, it can actually carry some evidential weight.
- 2:42Uchi Jayawardhena: Without going into too much detail, I think we can all relate to a couple of cases in the recent past where Elon Musk's tweets were used as evidence in some of the trials that he faced.
- 2:53David Shepherd: And we've seen situations in courts in England and Wales as well, where social media posts, even from third parties, have been used to evidence behaviour or intent or that specific events have occurred.
- 3:04David Shepherd: And so it's no longer unusual.
- 3:05David Shepherd: It's becoming a much more common evidential source, even in commercial disputes.
- 3:10Meghan Ryan: And it's not just the posts, it's the comments, replies, and you reshare to certain posts, um private messaging, which we usually don't have access to, um, stories that disappear overnight, um, and multimedia like images and video.
- 3:24Meghan Ryan: All this can be evidence connected to the social media posts themselves.
- 3:28Meghan Ryan: So, from a data perspective, it's a lot less structured than traditional sources.
- 3:32Roisin McCrory: One of the assumptions people often make is that if content is online, it should be easy to identify who created it.
- 3:38Roisin McCrory: But is that actually the case?
- 3:40Roisin McCrory: I mean, how difficult can it be to confidently link social media activity back to a specific individual?
- 3:47Uchi Jayawardhena: That's a great question, Roshin.
- 3:48Uchi Jayawardhena: I think you have some significant issues you're dealing with because people can have multiple accounts.
- 3:54Uchi Jayawardhena: You can have anonymous users, uh, you can have accounts which are not necessarily tied down to a specific individual or can't be clearly tied to an individual or an entity.
- 4:05Uchi Jayawardhena: And then on top of all this, you have posts that are made by AI bots.
- 4:09Uchi Jayawardhena: Now, this all complicates the collection of the day information as well as the interpretation of it.
- 4:16Uchi Jayawardhena: And identity doesn't become a legal issue or of its own or a legal question on its own.
- 4:22Uchi Jayawardhena: It impacts how the how reliable is the data set.
- 4:26Uchi Jayawardhena: It also impacts how can we confidentially link an account or an interaction to a specific person or an entity.
- 4:33Uchi Jayawardhena: And then whatever analysis that you subsequently do with this information becomes much weaker if you can't necessarily bring that confidence level to link it to a specific individual or an entity.
- 4:45Roisin McCrory: So before we even get to reviewing the evidence, we're already dating with questions around identity and ownership and access.
- 4:53Roisin McCrory: Um that feels like a good point to move into collection because that's where many of the really practical challenges begin.
- 4:59Roisin McCrory: Is that right, Megan?
- 5:01Meghan Ryan: Yes, collection is a really big hurdle when it comes to this type of data.
- 5:04Meghan Ryan: Um, unlike email and OneDrive data, where it's easy to export, Microsoft has built platforms so you can export your data in bulk.
- 5:12Meghan Ryan: Um, the social media platforms are exactly the opposite.
- 5:15Meghan Ryan: Um, the makers don't want you to export your data.
- 5:17Meghan Ryan: They want to keep everything on the platform.
- 5:20Meghan Ryan: Um, any APIs that can be plugged into them are limited.
- 5:23Meghan Ryan: Terms of service restrict what you can do.
- 5:26Meghan Ryan: And then there's the issue of private accounts where you can only export the data from the person who actually owns it.
- 5:31Meghan Ryan: So then you're only able to export what's publicly available, which might not be the evidence that you need.
- 5:36Meghan Ryan: So in many cases, collecting the data is harder than traditional data sources.
- 5:40Uchi Jayawardhena: And Megan, if I were to just to add to that, uh, then there's of course the technical and the jurisdictional layer on top of all this.
- 5:48Uchi Jayawardhena: The tools that you actually want to use can sometimes be used in one region, but can't be used in another.
- 5:54Uchi Jayawardhena: And as a result, your approach has to adapt depending on where this data sits.
- 5:59Uchi Jayawardhena: And from an analytics perspective, the way the data is collected becomes really important because you need to be reliable, you need to have the ability to rely on this authenticity of the data.
- 6:12Uchi Jayawardhena: And that becomes critical simply because if you don't have the right tools or the collection methods are inconsistent, you have issues with the data you collect.
- 6:23Uchi Jayawardhena: It can be inconsistent, it can be incomplete, and if you have poorly documented data, you can where you have lost metadata, it tends to impact the relationships that you're trying to build, it impacts the context that you're trying to understand.
- 6:38Uchi Jayawardhena: So this all complicates matters.
- 6:42Roisin McCrory: So what I'm hearing is that collection isn't difficult just because of one issue.
- 6:46Roisin McCrory: It's difficult because of these technical, legal, and operational challenges are all in to connect it.
- 6:53David Shepherd: Yeah, exactly, Roshin.
- 6:54David Shepherd: And I think there's a key distinction between social media data, which is public, and social media data, which is private content.
- 7:01David Shepherd: So public data may be accessible, and we've we've discussed briefly some of the challenges that one would still face in collecting that, but private content usually is not uh accessible.
- 7:11David Shepherd: And an increasing volume of social media content is private, so is, or at least limited to certain other users, so friends or contacts.
- 7:19David Shepherd: And that material, even though it's private, may still be crucial evidence, and if so, that's where your legal strategy is going to come in in order to get hold of that.
- 7:29David Shepherd: So if you and your client can't access uh certain social media data directly, you may need to access it through the traditional disclosure process in litigation.
- 7:37David Shepherd: So, for example, by ensuring that the right custodians are included, so individuals who uh have have access to and participate in that social media content and therefore have control over it for disclosure purposes.
- 7:49David Shepherd: And that's particularly important for private social media, so private groups or one-to-one messages to that particular custodian or otherwise social media which is anonymous, so where it's um posted on a username that's not connected specifically to a person of interest, it may be very cumbersome for you to work out which social media uh accounts are linked to that custodian to that person, but the disclosure process can reveal that information to you.
- 8:14David Shepherd: So that can offer a route into what is otherwise inaccessible information, inaccessible uh material.
- 8:19Roisin McCrory: In some cases, you might strongly suspect relevant information exists, but you're dependent on another party to preserve and disclose it.
- 8:27David Shepherd: And that's where you're going to need to look really carefully at your counterparty's disclosure in order to ensure that they are preserving what you want and they are collecting the social media content you want, and indeed telling you about the social media accounts or content that exists that you might be interested in.
- 8:42David Shepherd: And if the counterparty themselves can't access it, it may be necessary to consider whether a third party might voluntarily provide it, or otherwise even consider a third party disclosure application.
- 8:52Meghan Ryan: And that's where screenshots come in again, which really amplify this risk.
- 8:55Meghan Ryan: Um, when you just receive a screenshot, you can't verify completeness or authenticity without proper collection.
- 9:01Roisin McCrory: And another trend I think that must be really changing the landscape is the raise of disappearing messages and ephemeral content.
- 9:09Roisin McCrory: How does that affect the obligations around preservation and disclosure?
- 9:14David Shepherd: Yeah, really good question, Rasheen.
- 9:16David Shepherd: So I think the key issue there is timing.
- 9:18David Shepherd: So before the litigation is within reasonable contemplation, I think disappearing messages or content will be a matter for explanation for a party.
- 9:27David Shepherd: And then the usual approach will apply in terms of uh documents that used to exist and no longer do exist and were lost because of the document preservation policies that were in place, in this case, having disappearing messages on social media.
- 9:42David Shepherd: But once litigation is within reasonable contemplation and a party is under document preservation obligations, continuing to use such features would likely mean that they're in breach of those document preservation obligations.
- 9:54David Shepherd: And so that's an important point to bear in mind for our own disclosure, to ensure that any such features are turned off, and to consider for the other side's disclosure and whether there might be such a feature and when best to raise with them their their use of that.
- 10:08Uchi Jayawardhena: So, David, basically what you're saying is that something that's actually normal day-to-day communication between people can actually become a legal risk very, very quickly.
- 10:18David Shepherd: Yeah, exactly.
- 10:18David Shepherd: And if that content is then gone, if that's lost, that'd be left with gaps in the disclosure, which makes it harder potentially to prove or to disprove a contested issue with each side arguing uh before the court as to what they should infer to, in effect, plug that gap.
- 10:33Meghan Ryan: And it all depends on the platform as well.
- 10:35Meghan Ryan: Um, some platforms, even though it's disappearing messages, they might keep an archive that you're able to collect.
- 10:40Meghan Ryan: Um, but sometimes when it's gone, it's gone.
- 10:42Meghan Ryan: So you really don't know what's recoverable when you're dealing with those type of messages.
- 10:46Roisin McCrory: So is it a case that it often comes down to the device itself?
- 10:51David Shepherd: Yes, devices can can often be important to access.
- 10:53David Shepherd: So, for example, if the messages are saved locally.
- 10:56David Shepherd: And and again, you know, if the if the data is lost or unavailable, the court will assess whether it genuinely has been, and and if so, what that means for the evidence before it.
- 11:05Roisin McCrory: And and that's quite different, David, from traditional corporate systems where information may be stored in multiple locations.
- 11:11Roisin McCrory: But in the social media world, the loss of a single device can have really significant consequences.
- 11:18Roisin McCrory: Um, assuming then that you've managed to identify, you've preserved and collected the content, the next challenge is actually making sense of it.
- 11:27Roisin McCrory: And I expect that can be just as complicated, Megan.
- 11:30Meghan Ryan: Yes, it's very complicated.
- 11:32Meghan Ryan: Um, it's just very difficult to review.
- 11:34Meghan Ryan: It's not like a traditional email where it's structured and you're very reliant on the text.
- 11:39Meghan Ryan: Social media posts are very high volume, short form, very context dependent.
- 11:45Meghan Ryan: They're also very repetitive if um individuals are just resharing the same posts over and over again.
- 11:51Meghan Ryan: Um and a lot of the communication isn't text-based.
- 11:53Meghan Ryan: Um, there's images, videos, gifts, um, emojis, and search terms won't capture those types of items.
- 12:00Meghan Ryan: So if you're reviewing this in a traditional e-discovery platform, it's hard to search through uh this type of data.
- 12:06Uchi Jayawardhena: And Megan, I think this is where AI comes in, but it also has its own limitations.
- 12:11Uchi Jayawardhena: Now you can use AI to help with transcriptions, uh, do some image analysis, but unfortunately, most of these tools are still built around text and therefore they have limitations on what they can actually perform.
- 12:27Uchi Jayawardhena: And I think most importantly is we are we are using AI tools to do transcriptions over the image analysis.
- 12:34Uchi Jayawardhena: I think we have to be really careful because the process has to be transparent enough that it's a defensible process that if it is challenged as part of a litigation method.
- 12:46Meghan Ryan: Yeah, that's right.
- 12:47Meghan Ryan: And you can't really rely on AI to fully interpret video or context quite yet.
- 12:52Meghan Ryan: Um, so you still need human review.
- 12:54Meghan Ryan: Um, there's also a language challenge here, because social media evolved quickly.
- 12:58Meghan Ryan: There's slang, abbreviations, memes, and just specific references that AI might not naturally pick up.
- 13:05Meghan Ryan: Um, so there's still a lot of human review that's required, even with advances in AI.
- 13:10Uchi Jayawardhena: And if I was just to add something more to this, is now you're dealing with AI generated content on top of all this.
- 13:17Uchi Jayawardhena: So this just adds another layer of complexity.
- 13:20Uchi Jayawardhena: And what's the interesting is that AI generated content is becoming increasingly difficult to actually detect that it's AI generated itself.
- 13:30David Shepherd: Yeah, and that's a really interesting challenge, usually, from a disclosure perspective, where it amplifies that that issue of the court working out whether the evidence that's been disclosed is itself genuine.
- 13:41David Shepherd: So for most electronic disclosure that's come from a sort of corporate system in recent years, that hasn't really been much of an issue, as one can look to the metadata to see whether or not, say, a particular email is authentic or not.
- 13:52David Shepherd: But with that rise in AI-generated social media content, that is likely going to create situations where there is a dispute as to whether a particular post is itself genuine.
- 14:01David Shepherd: So is the account actually used by the individual of interest, or is it a fake bot account?
- 14:07David Shepherd: Did they actually take the photo that's been posted?
- 14:09David Shepherd: Did they type the text that's been posted, or is that all AI generated?
- 14:13David Shepherd: As of before turning to consider what the evidence shows and what weight to attach to it, the parties and indeed the court may need to grapple with whether it is actually genuine.
- 14:33Roisin McCrory: So looking ahead then, do we think social media data is becoming an expected source of evidence rather than something that's exceptional or case-specific?
- 14:43David Shepherd: I think it depends upon the case, but I don't think it's exceptional.
- 14:46David Shepherd: I think there's an increasing expectation that it will at least be considered during the disclosure process.
- 14:52David Shepherd: And so the parties would turn their mind to whether any of the key witnesses custodians, you know, communicate on instant messenger apps or post about relevant matters on social media.
- 15:01David Shepherd: And if they do, then um I think that you'd need a reason why not to include that within the collection process for disclosure.
- 15:08David Shepherd: For content that's posted publicly, litigators should bear that in mind in terms of the risk of the other party locating and producing that and, for example, using it in cross-examination.
- 15:18David Shepherd: Uh, and again, in such circumstances, you know, a challenge is likely going to be around whether it is authentic, you know, particularly with the AI-generated content that we've we've been discussing.
- 15:27David Shepherd: And, you know, it may be that uh in future cases there's a bigger call on expert um evidence to consider whether particular disclosure is authentic or not.
- 15:37Uchi Jayawardhena: Thanks, David.
- 15:37Uchi Jayawardhena: And I think this is where we have another issue coming up with is the tools that we are using because the platforms are evolving so quickly, uh, the data formats are getting so complex, and uh what we are playing a cat and mouse game of as the tech moves, uh the detection tools are trying to improve, and it's an evolving process.
- 15:59Uchi Jayawardhena: And this also creates an interesting paradox because we are now using AI tools to actually detect AI generated content.
- 16:06Uchi Jayawardhena: While this is useful, I think it's really important that we understand two elements.
- 16:11Uchi Jayawardhena: One is we also need to understand the limitations of the detection tools that we are using, and the second part is of course making sure that the process that is being followed is an actually defensible process if it is scrutinized or challenged.
- 16:26Meghan Ryan: Yes, because even if once we're able to collect this data, it is really a question about how you review it at the scale and just to make sure it's uh defensible, just like you said, Uchi.
- 16:36Roisin McCrory: So listening to all of this, it feels as though every new way of communicating creates you know fresh challenges for legal and technology teams.
- 16:45Roisin McCrory: It it begs the question: is this really about social media or is it a broader story about technology evolving faster than our ability to preserve, analyze, and verify that information?
- 16:58David Shepherd: I think that's probably right, Rosheen.
- 17:00David Shepherd: Many of us are using an increasing number of platforms uh through in in work through which to communicate with each other and through which to create and store work.
- 17:09David Shepherd: And where the that may or may not be more efficient for our work, it is certainly going to create additional challenges for disclosure processes going forward.
- 17:17David Shepherd: And you know, that probably is a point for organizations to be bearing in mind when they're considering the platforms that they um offer their teams.
- 17:25David Shepherd: You know, do they want people communicating on multiple different instant messenger platforms?
- 17:30David Shepherd: Do they want um multiple different social media uh platforms to which they're communicating with with clients and counterparties and other third parties?
- 17:40David Shepherd: Um maybe, maybe they do, and they do need that for commercial reasons, but you know, there are going to be complications if any of that's ever litigated.
- 17:48Meghan Ryan: For me, the key point is that social media data forces us to think beyond traditional document review.
- 17:53Meghan Ryan: We need approaches that can deal with networks, context, multimedia, identity and uncertainty, and not just traditional keywords that we sublime on documents.
- 18:01Uchi Jayawardhena: And Megan, if I can just add to that, I always look at it from an analytics point of view.
- 18:05Uchi Jayawardhena: So this is where analytics can add some real value to it in making sure that we've gathered the relevant data and built the context relationships in the data that has been collected.
- 18:16Uchi Jayawardhena: But this is all dependent on having a solid foundation.
- 18:20Uchi Jayawardhena: If the foundation is not solid, then the analytics is not going to be very helpful in pursuing these matter.
- 18:28Roisin McCrory: For me, if there's one takeaway, it's that social media data is high value, but it's high risk and low certainty.
- 18:36Roisin McCrory: So the challenge isn't just access, it's around that authenticity and it's around trust.
- 18:42Roisin McCrory: And I think what I'm hearing as well is capability will always lag change, and that gap's probably the most visible here with social media content.
- 18:52Roisin McCrory: So there you have it.
- 18:53Roisin McCrory: Thank you to David, Uchi, and Megan for taking part in this final episode of Legal Tech Deciphered, Trust, Truth, and Hallucinations.
- 19:02Roisin McCrory: And thank you for listening as well.
- 19:04Roisin McCrory: We hope that you find the series insightful.
- 19:07David Shepherd: Thanks very much for having me on, Roshin.
- 19:08David Shepherd: It's been a really interesting discussion.
- 19:10David Shepherd: Absolutely.
- 19:11David Shepherd: Uh lots to think about, and thanks everyone.
- 19:14Meghan Ryan: Thanks, all.
- 19:14Meghan Ryan: I really enjoyed it.