The Lab Notebook
Actionable Insights from the Vanderbilt AI Law Lab | No. 1
Welcome to The Lab Notebook, VAILL’s field log for tracking how AI is reshaping legal work. Our goal: fewer hot takes, more meaningful signals. We’re designing each Notebook entry to reflect an industry in motion: Version Control highlights recent meaningful updates, Open Query surfaces the unknowns we should be debating, Network Effects highlights what’s compounding through shared practice, and System Status checks the guardrails and institutional infrastructure. We close with The Signal, one high-fidelity takeaway that suggests where the future is heading.
🔁 Version Control
This week’s updates aren’t just “new model, new feature.” They’re small, yet meaningful, shifts that push the industry from experimental v1 workflows toward v2 systems that actually work end-to-end.
Frontier models moved again (and they’re explicitly pitching “workflows,” not “chat”). OpenAI launched GPT-5.2 and is emphasizing improved “factuality” + longer context + multi-step execution. 🔗
Anthropic pushed “agents + computer use” harder. Claude Opus 4.5 positions itself as a step toward agentic work (and not just writing). 🔗
Legal AI “workflow integration” is the new product headline. LexisNexis announced a next-gen Protégé positioned as an integrated workflow across legal content + customer docs + web insights. 🔗
Courts are treating “AI literacy” as infrastructure. Thomson Reuters is framing AI literacy as role-specific capability-building for courts (a shift from “policy memo” to “operational competency”). 🔗
🧐 Open Query
These are the questions that keep showing up in practice, but don’t have stable answers yet. Consider them prompts to run against your own experience (and your network’s) before the next wave hardens into “best practices.”
If AI flattens the BigLaw pyramid, what replaces apprenticeship? A Reuters piece flags the emerging “cylinder” model and the training/billing model stress it creates. 🔗
What’s the new malpractice standard for hallucinations—especially when the error is “plausible”? Courts keep surfacing hallucinated citations; the story is evolving from “lol” to repeating “failure mode with sanctions.” 🔗
Who governs AI in the U.S.: states, the feds, or . . . the courts? A brand-new White House executive action frames state AI laws as “obstruction,” teeing up federalism fights (and legal challenges). 🔗
Is “safety” about model behavior, or the incentive stack around deployment? State AGs are pressing major AI companies for stronger safeguards (notably around vulnerable users). 🔗
What’s the hidden environmental/infra layer of AI policy? EU discussion of fast-tracking AI datacentres (“gigafactories”) by easing permitting rules raises a collision between “AI competitiveness” and other regulatory regimes. 🔗
🌐 Network Effects
The advantage is compounding in the places that learn together. This section tracks where adoption, shared playbooks, and organizational effort are creating outsized returns—because the network makes each node smarter.
Innovation features as an internal role, not a vendor purchase. The FT’s intrapreneur feature (and its Innovative Lawyers awards coverage) spotlights partners/operators building in-house tools + governance + training as a repeatable pattern. 🔗
Adoption reporting is getting more granular (and more useful?). Coverage digging into legal AI adoption (including Harvey/RSGI discussion) suggests 2025 may be considered “the year adoption got real,” at least among tracked customer bases. 🔗
Survey data point to the “small firm lead” dynamic. One industry report’s respondent pool skews heavily toward smaller firms (e.g., ~66% in the 10–50 employee band), which can shape who moves fastest and why. 🔗
🚦 System Status
Call this the OS health check: what’s changing in the rules, guardrails, and institutional infrastructure that keeps the profession stable. Policies, ethics guidance, and evidentiary shifts are the early warning lights for where risk (as well as opportunity) moves next.
Courts are moving from “don’t do dumb stuff” to formal AI policies (with training + tool restrictions). New York State Unified Court System issued an Interim Policy on the Use of AI that sets guardrails for judges + staff (including security/confidentiality constraints and responsible-use expectations). 🔗
Evidence rules are starting to treat “deepfakes” as more than just a tech problem. The U.S. Courts’ Evidence Rules Committee materials discuss proposals aimed at handling authentication/evidentiary issues raised by deepfakes and synthetic media. 🔗
State bar guidance is getting more specific (competence, diligence, confidentiality, candor, supervision, and fees, mapped to RPCs). Washington State Bar’s 2025 advisory opinion walks through duties under multiple RPCs when lawyers use AI tools. 🔗
Billing + fees = the next ethics flashpoint. Virginia’s bar moved forward a proposed ethics opinion focused on the reasonableness of fees when lawyers use time-saving generative AI tools, signaling where disputes may emerge (and what bars expect). 🔗
📡 The Signal
The legal industry isn’t just upgrading models. Instead, we’re witnessing a quiet upgrade in the operating system of law. As AI shifts from “help me write” to “help me run the workflow,” the profession is responding with infrastructure: court AI policies, evidence-rule modernization for synthetic media, and bar guidance that extends to fees and supervision. The signal isn’t that tools are getting better (they probably will). It’s that the institutional layer (training, governance, accountability, legitimacy) is finally being rewritten to keep up.
We can “rewrite the rules to keep up”, but frankly, that’s not enough. We have to rewire the culture of law.
The problem is not the practice of law, it’s the business of law. Unless and until the culture shifts away from protection and income redistribution to the intermediaries between those that need legal services and that which creates the need (aka lawyers and #LawLand) and moves to a truly customer focused delivery ecosystem, then we’re really just moving the deck chairs on the Titanic. Another Substack commentator in this space has pondered whether Clayton Christensen’s theory of disruptive innovation applies in law.
I submit it’s not a theory, it’s a law. It’s only a question of timing. To date LawLand has successfully resisted disruptive innovation and has largely maintained the status quo. This has resulted in perpetuation of bespoke, highly curated, individualistic complexification of delivery systems that create ever increasing activity to generates revenue for the lawyer as opposed to serving the customer. In other words, the legal delivery system is focused primarily on the creation and perpetuation of disputes so that those most interested in answering interesting questions of law — a classic narcissistic intermediary — exact a intermediation tax from those that actually need the service/product. This in turn results in massive unmet latent demand as the true customers are simply priced out of the market for the services they actually need. The tools emerging today provide for P3 change (Processification | Platformization | Productization) for the vast majority of the legal needs by those accessing the delivery system today as well as that latent unmet demand. But it takes massive cultural shift in LawLand to get there — and I for one am not convinced that brownfield reclamation is possible. The massive active and passive resistance to cultural change is in the way.
However, disruptive innovation will take place and those clinging to the past delivery system will change or die — my money is on the latter — as it has been for decades. The disruption is inevitable — however the timing remains uncertain. The defenders of LawLand are legion — the walls of the citadel are thick, the moats are deep. One doesn’t attack a citadel of this type frontally, no, you surround it,bypass it and starve it. As such, greenfield providers — true disruptors — are likely the ultimate answer. The disruptors need only proof of concept for the customers to defect, to vote with their fees. But that also requires the buyers — most of whom are from the same guild and suffer from the same professional narcissism to change themselves and accept that they are agents of the customer and that it is their fiduciary duty to demand disruption.
As has been the case for some time — the answer to the problem is not difficult, but it is hard, very very hard because those with the power to change struggle to get there.