All the intelligence that's fit to argue.
Illinois moved decisively into the national AI governance vacuum last week when the Illinois House of Representatives passed Senate Bill 315 — the Artificial Intelligence Safety Measures Act — by a unanimous 110-0 vote on May 27, sending the most aggressive frontier AI oversight law in the country to Governor J.B. Pritzker, who has publicly pledged to sign it. The Senate had cleared the bill 52–5. If enacted, SB 315 will require large frontier AI developers — companies whose global affiliates report more than $500 million in annual revenue — to publish annual safety and capability risk reports, disclose critical safety incidents within 72 hours, and, in a first for any U.S. jurisdiction, submit to mandatory annual independent third-party audits of their safety practices. Supporters of OpenAI and Anthropic backed the bill; major AI trade associations opposed it. Illinois would become the third state to regulate frontier models, joining New York's RAISE Act and California's SB 53 — but SB 315 goes further by making outside auditor verification a legal requirement rather than a voluntary commitment. Key compliance provisions take effect January 1, 2028, with Illinois Attorney General enforcement authority and civil penalties. Legal commentators are already invoking the "Sacramento Effect": because frontier developers are unlikely to build Illinois-only audit pipelines, SB 315's requirements may functionally become the national baseline regardless of federal inaction.
Reporter's Take: For WC/PI practitioners, the practical import is indirect but real — the AI tools shaping claim valuations, IME scheduling, medical record review, and settlement analytics are built on the very frontier models this law will audit. When those systems are externally verified for safety and catastrophic-risk controls, defense counsel and carrier AI deployments will carry a new layer of accountability that plaintiff attorneys can invoke in discovery and Daubert challenges alike.
The world's highest-grossing law firm announced a multi-year agreement with Palantir Technologies on June 4 to build a proprietary AI platform for private equity fund formation — the first deployment under Kirkland's $500 million AI investment strategy, initially serving 1,000+ attorneys in its investment fund group. The platform embeds senior partner institutional knowledge into a scalable AI system for drafting fund documents, managing side letters, and tracking investor compliance obligations across global fundraises. Why it matters: When the Am Law 1 firm spends half a billion dollars operationalizing AI knowledge transfer from senior partners to platform, it reframes "AI strategy" as infrastructure — not software subscription. WC/PI boutiques should watch whether this institutional-knowledge-embedding model migrates to defense firms handling high-volume carrier relationships.
Source: Business Wire / Private Equity WireSuperlegal, founded by the former founders of LawGeex, launched June 3 as what it bills as the first AI law firm in the United States authorized to practice law — operating under the Utah Supreme Court's Legal Services Innovation Sandbox with non-lawyer ownership permitted. Targeting the $2.1 trillion construction industry first, Superlegal reviews and redlines commercial contracts in under 24 hours for as low as $117 per contract, with a licensed attorney signing off on each review. It has partnered with the Associated General Contractors of America for member distribution. Why it matters: Utah's sandbox is the regulatory beachhead; the next logical markets are high-volume, document-intensive practices — including subcontractor injury claims, construction-site WC defense, and general contractor indemnity disputes where contract review volume is crushing and margins are thin.
Source: GlobeNewswire / Artificial LawyerArtificial Lawyer published a deep-dive this week flagging a systemic crisis quietly reshaping every legal AI vendor's economics: token costs are rising faster than capability improvements can offset them, particularly for agentic workloads that consume orders-of-magnitude more tokens than single-query tools. Harvey CEO Winston Weinberg acknowledged the firm is routing tasks to cheaper models via Factory to manage infrastructure costs. Flank co-founder Jake Jones noted pricing is already shifting away from per-seat licenses and toward displacement-based models — "priced on the displacement, not the licence." Why it matters: Law firms mid-contract on flat-rate AI subscriptions need to understand whether those deals survive vendor repricing pressure as agentic usage balloons costs by 5–30x.
Source: Artificial LawyerAlphaSense, the AI-powered market intelligence platform, closed a $350 million round on June 3 valuing the company at $7.5 billion — nearly double its prior $4 billion mark — with $600 million in ARR exceeded in Q1 2026. The round was led by Vitruvian Partners, Accenture Ventures, and J.P. Morgan Asset Management. Accenture becomes its first strategic channel partner. The company unveiled SuperAnalyst, an always-on AI agent for automating financial and strategic workflow tasks. AlphaSense now serves 70%+ of S&P 500 companies and nearly all major financial institutions. Why it matters: AlphaSense's document intelligence architecture — deep cross-source search with structured reasoning — is precisely the kind of AI infrastructure carriers use to build settlement analytics. Watch for insurance vertical expansion.
Source: AlphaSense Press Release / National Law ReviewMedRisk's 2026 Industry Trends Report documents the sector's central paradox: claim frequency is declining while severity trends persistently higher, as injured populations age and comorbidities increase. Meanwhile, 55% of adjusters in 2026 report too many claims to manage — up from 40% in 2025 — per Healthesystems' annual survey. Carriers and TPAs have stopped debating AI adoption and are now grappling with how to scale it responsibly. The report notes that time from injury to first conservative treatment for lumbar spine claims fell 50% between 2021 and 2025; time to advanced imaging fell 53%. Why it matters: Faster care access compresses the medical causation timeline — which shifts the battleground in contested WC claims from "did treatment occur" to "was the treatment necessary and correctly attributed."
Source: MedRisk / Risk & InsuranceA demonstration at Legalweek 2026 in March that continues to circulate in legal tech circles: a live AI agent given a single litigation hold notice autonomously identified custodians, mapped data sources, drafted preservation letters, and scheduled collection workflows — in under four minutes. The room, per observers, went quiet for the wrong reason. Not awe. Unease. Because what they were watching wasn't a faster tool. It was a different kind of worker: one that doesn't bill hours, doesn't need training, and doesn't skip steps under deadline pressure.
The major legal tech vendors — Thomson Reuters CoCounsel Legal (now 1 million+ professional users), LexisNexis Protégé, DISCO, Harvey, and Epiq — all shipped agentic capabilities in Q1 2026. The architecture varies: some use orchestrator-plus-specialist agent models (LexisNexis deploys a four-agent team: orchestrator, legal research, web search, and document agents); others embed agentic planning at the model layer (Thomson Reuters integrates the Claude Agent SDK directly to enable "one conversation" multi-step research grounded in Westlaw and Practical Law).
For WC/PI builders and firms: Agentic litigation hold, discovery mapping, and custodian identification are table stakes now. The next deployment frontier in WC/PI is agentic claims intake triage — an agent that receives a new injury report, queries prior WC history, flags potential fraud indicators, identifies applicable state fee schedules, and surfaces the top three IME candidates — before a human adjuster ever opens the file. The carriers building this first will have a 30-day claims-handling head start on every contested claim.
MedRisk's 2026 Industry Outlook confirms what practitioners on the ground already sense: AI in workers' compensation has crossed from experimentation to operational embedding. Carriers and TPAs are deploying AI for intake triage, care coordination, utilization management, and outcome prediction as part of their standard claims workflow — not as add-on pilots. The key operational finding: early intervention via AI-identified risk flags is now the primary driver of improved outcomes, not cost containment per se. Time to first conservative treatment for lumbar spine injuries dropped 50% over five years; time to imaging dropped 53%. For plaintiff counsel, this creates a two-sided discovery opportunity: when outcomes are good, carriers will credit the AI system; when outcomes are poor (delayed treatment, wrong care pathway, under-reserved severity), the same AI audit trail becomes evidence of systemic failure.
Healthesystems' 2026 survey found that 55% of workers' comp adjusters report too many claims to manage — a 15-percentage-point jump in one year. Industry voices are framing this as an argument for more AI augmentation, but for litigators it signals something different: the human oversight layer in AI-assisted claims decisions is thinning. When an AI triage system flags a claim for fast-track resolution and an overworked adjuster rubber-stamps the recommendation, the medical necessity review that should have happened — the IME, the functional capacity evaluation, the independent causation analysis — may not occur. That's a compensability fight waiting to happen, and the adjuster's workload documentation becomes relevant to whether the carrier's reliance on AI constitutes unreasonable claims handling.
Faster access to advanced imaging (MRI, CT) — now averaging 53% shorter lag than 2021 — changes the evidentiary landscape in WC/PI causation disputes. When imaging happens within days of injury rather than weeks, the pre-litigation medical record is more complete, pre-existing conditions are documented earlier, and the plaintiff's treating physician's causation opinion is harder to attack on the grounds that imaging was delayed or incomplete. Defense IME physicians who previously relied on delayed imaging to argue insufficient diagnostic workup face a stronger initial medical record. Expect AI-accelerated imaging timelines to make early causation stipulations more common — and early causation fights sharper.
MedRisk's 2026 annual trends report is the most data-dense public document on the current state of AI in workers' compensation claims operations — covering adjuster capacity, care timeline improvements, AI deployment maturity, and claims complexity drivers. It is written for the carrier/TPA operational audience but contains exactly the kind of baseline data that plaintiff WC counsel needs to understand what the other side's AI systems are designed to do and how performance is being measured. The report's finding that claim frequency is falling while severity rises — driven by aging workers and comorbidities — directly tracks with the contested-claim scenarios where WC attorneys earn their fees.
Every AI-assisted claims decision — reserve-setting, triage flagging, care authorization, litigation risk scoring — now generates a documented audit trail that didn't exist under purely human adjuster models. This is the most underappreciated structural shift in WC litigation: not that AI is making decisions, but that it's making documented decisions. Discovery practice in contested WC claims needs to evolve immediately to request AI system documentation, training data vintage, model version logs, and output records. The carrier's AI platform is no longer just an operations tool — it's a discovery target.
SB 315 joining HB 3773 (Subpart J employment AI rules) and the NICU Leave Act in the same legislative session signals that Illinois is systematically building the most comprehensive AI-adjacent employment and safety law framework in the country. For WC/PI practitioners licensed in Illinois, this is a competitive edge: you are operating in the most legally sophisticated AI governance environment outside California. When federal rules eventually arrive — and they will — they are likely to converge on the Illinois/California/New York framework. Learning this framework now is not compliance overhead. It's a five-year head start.
$117 per contract, 24-hour turnaround, attorney in the loop, Utah sandbox authorization. Superlegal's construction industry launch is a proof of concept for AI-native legal service delivery in any high-volume, document-intensive practice area. WC defense — where carriers farm out hundreds of subcontracting agreements, independent contractor MSAs, and settlement releases — is a natural second market. The question isn't whether this model comes for volume WC work. It's which incumbent firm builds the equivalent first.