Since Open AI’s launch of ChatGPT in 2022 at least one prediction annually has declared that the coming year would be the year that the eDiscovery, digital forensics, and cybersecurity markets would go all-in on Generative AI (Gen AI). In 2023, we saw the introduction of Gen AI-researched legal briefs and the first sanctions for not checking Gen AI-assisted citations. While
Mata v. Avianca may or may not have been the first Gen AI-assisted pleading, it certainly was not the last. 2024 was also a year where we saw a flurry of Gen AI-based applications for the legal services and cybersecurity industries hit the market. An often-cited study on Gen AI and cybersecurity noted that in 2023 to mid-2024, “[s]eventy-five percent of security professionals said they have seen an uptick in attacks over the past year, with 85% attributing the rise to bad actors using generative AI.” Therefore, with another January and a fresh set of calendar pages in front of us, this is probably the time to make our predictions – is 2025 the year we finally go all-in on Gen AI?
eDiscovery: No, but Seismic Changes
Our prediction for the eDiscovery market is that 2025 will continue to see the introduction, testing, and refinement of Gen AI tools for eDiscovery functions, but not a complete adoption of Gen AI tools…yet. However, what the increase in Gen AI tools will mean for the eDiscovery services sector is some change to how things are done. Services that are traditionally priced on a per hour basis, such as document review, may become increasingly part of the realm of Gen AI-enabled tools. This doesn’t mean that the services of eDiscovery professionals will go away, but rather, expertise will be an even more valuable asset as the need increases for sophisticated and analytical prompt engineering and careful quality control of Large Language Model (LLM) outputs. As law firms and service providers continue to seek out the right Gen AI tools, the dollars allocated to those tools will increase, shifting costs to new budget categories. While fees for automated services may see downward pressure, “technology fees” to help firms recoup the significant investment, configuration, and training costs may come online or even increase.
A newer cost that is being introduced by service providers in many industry sectors is “token” or “credit” pricing. Many Gen AI tools require the purchase of tokens or credits that are consumed at various rates as the tools perform functions such as search, summarize, analyze, or create. Tokens and credits, when required, are over and above other license and storage fees charged by developers and are necessary to access the full range of tool functionality. eDiscovery service providers should know in advance if the tools they use have unlimited LLM computing usage and storage (and hence, a higher initial license cost) or have a structured token or credit pricing plan. Pricing models for tokens and credits are in a fluid stage and may change as the market develops.
Digital Forensics: We’re Getting Closer
Our digital forensics predictions for 2025 are that deepfake detection will improve and we may even see some consensus on watermarking, as well as increased reliability for image and video identification, all of which will advance digital forensics investigations immensely. Some real hope exists for detection technology getting better at ferreting out deepfakes in 2025.
Cybersecurity: The good, the bad, and wow
2025 could be a huge year for cybersecurity on all fronts. By the end of 2024, cyberattacks, including ransomware and business email compromises, were on the rise, in some part due to Gen AI making it easier to create more credible social engineering schemes. 2025 will see this ugly trend continue with increasing sophistication. We also look forward to seeing increased efficacy in vulnerability assessment tools helping to identify gaps in network security as well as remediation tools that provide real-time response to hacking attempts. Data security, especially where LLMs and Gen AI are involved, will be another area in which 2025 should see not only exposure, but also progress. The wow prediction we’re making for cybersecurity in 2025: Nation-state actors and proxies will step up their efforts to expose critical vulnerabilities in the US’ infrastructure, manufacturing sector, and supply chain. Our advice: act now to harden network security, remove unnecessary data from vulnerable locations, and implement regular training. The human component of cybersecurity will remain the most important.
2025 is already a big year for both the promise of new technologies and the escalation of threats we’ve seen before. For some of our predictions, we’re looking forward to ushering in an age where all sectors undergo exciting transformations with the help of AI. No matter what this year brings, Digital Mountain is here to guide you through another year of eDiscovery, Digital Forensics, and Cybersecurity adventures.
