Nine questions insurance executives should be asking about indemnity protection, A.I. governance, and control of litigation intelligence.

Commercial liability insurers face a troubling convergence. According to Verisk’s June 2026 Executive Insights Report, general liability claim severity increased 14% in 2025, far outpacing general inflation. S&P Global Market Intelligence reports that other liability occurrence insurers experienced $7.3 billion of adverse reserve development during the year.

Meanwhile, plaintiff law firms are deploying increasingly sophisticated A.I. to maximize settlement recoveries. That matters because, according to the 2026 CLM Litigation Management Study, approximately 97% of non-dismissed litigated claims resolve through negotiated settlements. Defense firms are adopting AI too, but primarily to improve research, drafting, and attorney productivity. Insurers bear the indemnity exposure, yet have largely left decisions about defense-side A.I. to their independent law firms.

How plaintiff firms got there is instructive. Few built their own A.I. They adopted shared, specialized, purpose-built platforms, and that model points to a solution carriers can bring to their own defense panels.

Without that kind of shared approach on the defense side, insurers face three potential losses of control: over settlement economics, over the security and governance of claim information, and over the intelligence generated across the defense panel.

Nine questions reveal these exposures.

Loss of Indemnity Control

1. Severities are soaring. How much of that is a negotiation problem?

Social inflation, litigation financing, and jury behavior all push severity upward. But Sedgwick’s 2026 Liability Observations Report highlights that settlement increases in general liability are outpacing verdict value increases. If settlements are rising faster than jury awards, part of the problem is happening at the negotiating table, which is the one place carriers can influence directly.

Plaintiff firms are industrializing AI-supported valuation, demand preparation, and negotiation. Carriers should set an equally rigorous standard across their panels: timely liability and damages evaluations, quantified trial exposure, bargaining-zone analysis, disciplined offer strategy, and persuasive written advocacy. Without a carrier-level standard, practice naturally varies by attorney and firm, and no one can measure it consistently across the portfolio.

2. Plaintiff firms have AI to maximize settlements. How are we equipping our defense firms to counter that?

Platforms such as EvenUp give plaintiff firms demand packages, settlement valuation, negotiation preparation, and historical settlement intelligence. Those tools exist to raise settlement amounts. Defense firms have invested sensibly in research, drafting, and summarization tools that make legal work more efficient. Those tools were simply not designed to improve results at the table.

Consider how plaintiff firms acquired that capability. Rather than building it, they subscribed to a specialized platform whose development cost and data are spread across many firms and matters. A defense firm handling one slice of a carrier’s docket cannot realistically build the defense equivalent alone, nor should it have to.

3. Which attorneys are negotiating our litigated claims, and how do we measure and support their success?

The 2024 CLM Defense Counsel Study found that only 8% of surveyed defense attorneys had received formal negotiation training. A Suite 200 Solutions Industry Snapshot revealed that offers are being conveyed by defense attorneys on roughly 60% of files. The 2026 CLM study found that 82% of responding claims organizations do not track who specifically extends settlement offers.

Put those together. Defense counsel convey most offers, few have had formal negotiation training, and most carriers cannot say which attorney or claim professional made which offer. The gap belongs to the industry rather than to individual lawyers. Carrier-supported training would help. So would a shared workflow in which every evaluation, offer, and counteroffer is recorded the same way, so that performance becomes visible as a byproduct of the work.

Loss of Governance Control

4. Are we still exchanging sensitive claim files with defense counsel by email, and who controls that exchange?

Litigation runs on medical records, expert reports, privileged communications, and defense strategy. Secure exchange systems are common, yet email with sensitive attachments persists in litigation workflows. AI adds another hop: claim documents now move from the firm’s inbox into analytical tools the carrier may never have seen.

John Burge
Operating Executive
SigmaSight

The real question is control. Once a file crosses into a defense firm and that firm’s technology vendors, who governs access, permissions, traceability, and retention? Common, secure exchange infrastructure can strengthen both carrier oversight and law firm operations.

5. Would the systems our defense firms use to process our claims pass our own cybersecurity review?

Carriers invest heavily in cybersecurity and vendor oversight. Their defense firms operate on far smaller technology budgets, and A.I. multiplies their dependencies: cloud environments, model providers, data retention terms, and access controls.

Auditing every panel firm’s A.I. stack is impractical. Expecting every firm to build enterprise-grade infrastructure is unrealistic. The workable path reverses the review. The carrier’s security team vets a small number of solutions once, approves them, and makes them available to the panel. Each firm then works inside an environment the carrier already trusts, without having to prove its own infrastructure carrier by carrier.

6. Which A.I. systems are shaping our negotiation posture and written advocacy, and who approved them?

A.I. is moving into substantive litigation functions such as analyzing liability, assessing damages, recommending negotiation positions, and drafting persuasive arguments. That raises questions of reliability, privilege, and permissible data use.

Carriers need not dictate every research or productivity tool their defense firms use. However, carriers will increasingly need to explain how AI contributed to a settlement position, who approved the underlying models and data protections, and whether the recommendations trace to reliable evidence.

Pre-approving those capabilities for panel use is a manageable task. Second-guessing each firm’s technology choices after the fact serves no one.

Loss of Intelligence Control

7. We pay for thousands of litigation settlements. Who’s getting smarter from them?

Defense firms can use AI to learn from matters spanning many insurance clients. But each firm sees only part of any one carrier’s portfolio. Carriers need aggregation in the opposite direction, across every firm handling their claims.

Imagine standardized liability assessments, trial-risk estimates, valuations, demand histories, offer packages, and outcomes captured across the entire panel. Carriers could see what works by venue, plaintiff firm, and defense attorney, and panel firms could benchmark their own results against the same data. When panel firms work on a common, carrier-approved platform, that data arrives in comparable form. Contractual rights and technical safeguards can enable carrier-specific aggregation while respecting work product and professional obligations.

8. Did our reserves see the exposure our defense lawyers saw?

Adverse development on the 2025 scale raises an uncomfortable question. When defense counsel uncovers new liability evidence or revises expected verdict exposure, how quickly does that reach claims management and reserving? Does it arrive as a structured, quantitative assessment, or in a narrative report format that was never designed to feed reserving?

Not every reserve change is foreseeable. But litigation produces a continuous stream of information that could flag deteriorating exposures earlier and, in aggregate, inform actuarial analysis, underwriting, and pricing.

9. Fraud and legal system abuse cross carrier lines. Why doesn’t our litigation intelligence?

Suspicious billing, recurring providers and experts, and coordinated litigation activity do not stop at one insurer’s boundary, yet each carrier sees only isolated claims. The pattern may emerge only when information from multiple insurers and public litigation records is analyzed together through structured evidence analysis, entity resolution, and controlled sharing.

The Answer: Vet the Platform, Then Set the Standard

The nine questions lead to one conclusion. Insurers cannot reasonably expect each independent defense firm to develop, on its own, the technology, security controls, and negotiation standards that protect indemnity, particularly when much of the economic benefit flows to the carrier and its insureds.

Nor do carriers need to build that capability themselves. They can do what they already do with every other critical business partner: review, approve, and authorize.

Carriers should pre-vet and pre-authorize a small set of defense-oriented negotiation and intelligence platforms, then make them available across their panels.

The carrier’s security and A.I. governance teams review a platform once. Claims leadership works with panel firms to set expectations for early evaluation, written advocacy, training, and outcome reporting, and the platform supports those expectations by design. Platform costs are authorized to individual claim files, and data rights are settled by contract up front so the intelligence reaches the carrier in usable form.

That one decision restores control in all three areas. Negotiation becomes consistent and measurable. Claim data stays inside an environment the carrier has approved. Every matter adds to the carrier’s own intelligence.

For defense firms, the model removes a burden rather than adding one. They avoid building and securing costly technology alone, platform costs are authorized to the file rather than absorbed as overhead, and one carrier review settles the security questions. They gain analytics few could build themselves and a measurable record of the value they deliver, while keeping their professional judgment and freedom to use other properly governed tools.

No industry standard or committee is required. A carrier can start with one or two trusted firms, measure the results, and expand across its panel.

Plaintiff firms succeeded by sharing specialized, purpose-built platforms rather than each building its own. Carriers and their defense partners can do the same.

Carriers pay for the settlements. They should be getting smarter from every one. &