Complex Discovery

💡 Market Intelligence: review's quiet retreat, collection's quiet surge - the eDiscovery task arc, 2012 to 2030

🗓️ Across 18 years, the composition of where eDiscovery dollars get spent across the three core tasks has fundamentally rebalanced. The RAND Corporation’s foundational 2012 study, Where the Money Goes, placed review at 73 percent of total task spend, processing at 19 percent, and collection at 8 percent. By 2025, the reconciled view places review at 62 percent, processing at 21 percent, and collection at 17 percent. By 2030, the forecast places review at 52 percent, processing at 23 percent, and collection at 25 percent. Review has lost 21 percentage points of share across the long horizon; collection has expanded its share over threefold.

⚖️ The pace of that rebalance is accelerating. The five-year share movement from 2025 to 2030 is nearly as large as the 13-year movement from 2012 to 2025. Two structural forces sustain the acceleration. AI-assisted review continues to compress per-document review labor faster than data growth can offset on the spend side – predictive coding now joined by generative-AI-assisted review and emerging agentic workflow features. Data-source proliferation continues to expand the collection scope – cloud collaboration platforms, mobile and ephemeral messaging, IoT, structured operational systems, and now generative AI prompt logs and model outputs each adding to the discoverable universe.

🔎 For cybersecurity, data privacy, regulatory compliance, and eDiscovery professionals, three observations follow. First, the task-share rebalance is structural rather than cyclical – the pattern has been visible since 2012 and has not reversed. Second, capital allocation and capability planning should anticipate the continued drift toward collection-heavy and processing-heavy task profiles. Third, the long-horizon view contextualizes near-term decisions: vendors and providers anchored primarily to review revenue should plan for continued share compression, while those with strong forensic and modern-collection capabilities are positioned for the fastest-growing absolute spend. The data volume context that follows this article will frame the demand-side trajectory in zettabytes that has driven this evolution, culminating in the consolidated 2025–2030 eDiscovery Marketplace Mashup.

📰 Read the complete article from ComplexDiscovery OÜ's industry research beat at https://complexd.blog/3RUFx3g.

🔎 Microsoft's first reasoning model arrives with a provenance pitch aimed at compliance teams

🖥️ Training-data provenance has become a productized sales argument in enterprise AI, and Microsoft moved early and explicitly to make it one. At Build 2026 in San Francisco on June 2, the company unveiled seven in-house MAI models led by MAI-Thinking-1, its first dedicated reasoning model, and paired the technical launch with a direct pitch to enterprise legal and compliance buyers. Microsoft’s public positioning is clean: commercially licensed data, no distillation from third-party models, and an enterprise-grade lineage general counsels can trust. The technical paper Microsoft published alongside the keynote is more nuanced: the corpus is “publicly available and licensed human-generated data” that includes a proprietary web crawl of approximately 1.2 trillion pages filtered to roughly 794 billion, a description analyst Simon Willison read as having “the same licensing problems as all of the other major LLMs.”

⚖️ For cybersecurity, information governance, eDiscovery, data privacy, and regulatory compliance professionals, the gap between the public positioning and the technical paper is the story.

👀 Watch whether Microsoft converts keynote language into contractual indemnification, whether early-adopter deployments produce auditable vertical benchmarks, and whether the marketing-versus-paper distinction holds up in procurement redlines.

📰 Read the complete article from ComplexDiscovery OÜ's artificial intelligence beat at https://complexd.blog/4ofvkuA.

🔎 Market Intelligence: where eDiscovery work gets bought - the delivery approach view, 2025 to 2030

⚖️ Quick read on where eDiscovery work is actually getting bought through 2030.
In 2025, corporations and governments bought about 72% of direct eDiscovery work — buying directly from vendors and service providers, without going through outside counsel. By 2030 the share is projected to ease to 67%. The in-house channel is the biggest, and will stay that way.

👀 But here's the shift worth watching:

+ Specialty service providers grow about 12% a year
+ Law firms grow about 10%
+ In-house teams grow about 6%

🖥️ AI is the underlying reason. AI-powered eDiscovery work scales economically through specialty providers that combine shared cloud infrastructure with operational depth in-house teams find expensive to staff and retain. Law firms are also gaining share again on the legally complicated stuff: AI regulation, cross-border investigations, complex strategy. Coordinating everything still matters.
The headline doesn't move much — in-house still dominates. But by 2030, more of the procurement is projected to be going through specialty providers and law firms.

📰 Read the complete article from ComplexDiscovery OÜ's industry research beat at https://complexd.blog/3S0GRl9.

💡 Glasswing widens: Anthropic puts Mythos inside power, water and hospital operators across more than 15 countries

🌏️ Anthropic on Tuesday expanded Project Glasswing beyond its roughly 50 initial partners, extending access to a new cohort of approximately 150 organizations in more than 15 countries. The restricted Claude Mythos Preview offensive-security model has already surfaced more than 10,000 high- or critical-severity vulnerabilities, according to Anthropic. The expansion lands one day after Anthropic’s confidential S-1 filing and pushes the frontier vulnerability-discovery tool deeper into power utilities, water authorities, hospitals, telecommunications carriers and hardware manufacturers, including organizations reported in Australia, Belgium, Canada, France, Germany, India, Italy, Japan, the Netherlands, New Zealand, South Korea, Spain, Sweden and Switzerland.

🔎 For cybersecurity, data privacy, regulatory compliance and eDiscovery professionals, the shift matters at three levels. Defenders inside Glasswing gain access to a restricted AI-enabled vulnerability-discovery capability at a time when Anthropic warns comparable models may become broadly available within six to 12 months. Counsel and information governance teams face a developing preservation and records-retention question around AI-generated vulnerability inventories, particularly in post-breach matters where Mythos-derived findings may become relevant. Cyber-insurance underwriters and vendor-diligence teams also gain a potential new control variable if Mythos-class access becomes a marker of advanced security posture. The same Tuesday, President Trump signed a narrowed AI security executive order establishing a voluntary 30-day pre-release review framework for covered frontier models and a Treasury-led AI cybersecurity clearinghouse.

🖥️ Watch the executive-order implementation deadlines, the NSA covered-model designation process, the Cyber Verification Program’s expansion, and OpenAI’s GPT-5.5-Cyber rollout.

📰 Read the complete article from ComplexDiscovery OÜ's cybersecurity beat at https://complexd.blog/4dM99Zh.