Insurance Industry IP Toolkit: Wordings, Rating Models, Data Sources, and Distribution

By Casey Scott McKay · 2026-04-09

An insurance product is a document, a model, and a distribution arrangement, and each of the three raises intellectual property questions that the sector has never resolved cleanly. This toolkit assembles the working material for practitioners advising carriers, brokers, managing general agents, and the technology vendors who serve them. It covers copyright in policy wordings and the bureau form system that supplies most of them, the trade secret position of rating models and the regulatory filings that undermine it, the third-party data licences that constrain product design, and the delegated authority arrangements where ownership of wordings, models, and customer relationships is most often left undefined. It closes with clause language, an authorities table, and the failures that recur.

IP and Technology > General IP | Toolkit | Published 9 April 2026 - Updated 29 July 2026 | Casey Scott McKay - marksy.us

Summary. An insurance product is a wording, a model, and a distribution arrangement. This toolkit covers copyright in policy language and the bureau forms most carriers rely on, the tension between rating model secrecy and rate filing disclosure, the third-party data licences that constrain what a product can price on, the delegated authority arrangements where ownership is routinely undefined, claims system and vendor rights, and the brand and filing layer that gates every launch.

Keywords: insurance IP · policy wording copyright · bureau forms · manuscript wordings · rating model secrecy · rate filing disclosure · third-party data licences · telematics · broker agreements · delegated authority · managing general agents · claims system rights · insurtech vendor terms · product filing · distribution marks

Start Here

Insurance is an information business that has never been comfortable treating its information as property, and three structural features explain almost every problem in the practice.

The product is a document, and most of the document is somebody else's. Standard policy forms produced by advisory and rating organisations are used across the market under licence. A carrier's "own" wording is frequently a bureau form with endorsements. Copyright in the base form belongs to the bureau; copyright in the manuscript endorsements may belong to the carrier, the broker who drafted them, or nobody in particular.

The pricing is a model, and the regulator requires it to be explained. Rate filings require actuarial justification, and the filed material may be publicly available. A model whose value depends on secrecy sits awkwardly with a regime that requires it to be described to a regulator and, in some states, to be discoverable by competitors through the filing record.

The distribution is delegated, and the delegation contract rarely addresses ownership. Managing general agents, programme administrators, and brokers develop wordings, build books, and accumulate customer data under agreements that specify commission and underwriting authority in detail and say nothing about who owns the product on termination.

Layer on the sector's data appetite — credit data, motor vehicle records, medical data, property characteristics, telematics, aerial imagery, catastrophe models — and the licence layer becomes the binding constraint on what a product can be.

Four questions organise the work.

Who owns the wording, and what may each party do with it?

Can the model be protected given what must be filed?

What do the data licences permit, and does the product exceed them?

And what happens to the wordings, the models, and the book when the distribution arrangement ends?

See The Wording Is the Product for the doctrinal treatment, Advising an Insurer or Broker for the sequence, and the Insurance IP Checklist for the working list.

Part one: copyright in policy wordings

Wordings are literary works and attract copyright as original expression under 17 U.S.C. § 102, subject to the ordinary limits.

Originality is thin but real. Feist Publications, Inc. v. Rural Telephone Service Co. requires a modest degree of creativity; policy language largely satisfies it, though a clause dictated entirely by statute does not.

The merger and scènes à faire limits bite hard. Where a concept can only be expressed in a small number of ways, expression merges with idea and protection fails — the analysis descends from Baker v. Selden and is applied to functional text routinely. Insurance drafting is full of language that is effectively dictated by the concept being insured.

Statutory and regulatory text is not protected, and mandated wordings incorporated into a policy carry no protection for the mandating language.

Bureau forms are licensed, not free. Advisory organisations produce standard forms, register the copyright, and license them to member and subscriber carriers. The licence terms govern reproduction, modification, and use by non-subscribers, and a carrier that leaves the arrangement needs to know what it may continue to use.

Manuscript wordings are the contested category. A bespoke wording drafted for a programme may have been written by the carrier's product team, by the broker, by the managing general agent, or by outside counsel. Absent a written assignment, the drafter owns it. Under 17 U.S.C. § 201 copyright vests in the author, and 17 U.S.C. § 204 requires a signed writing to transfer it. The work-made-for-hire route under 17 U.S.C. § 101 covers employees and only nine categories of commissioned work, none of which obviously fits a policy wording. See Who Owns the Work.

Registration matters for statutory damages and fees under 17 U.S.C. § 412 and 17 U.S.C. § 504, and Fourth Estate Public Benefit Corp. v. Wall-Street.com, LLC requires registration to have completed before suit. Carriers register wordings rarely and should register the ones that matter.

Enforcement is uncommon and not unknown. The sector's norm is imitation, and a carrier that copies a competitor's wording verbatim is doing something the market broadly tolerates until somebody decides otherwise.

Practice note. The practical value of wording copyright is defensive and transactional rather than litigious: it establishes what a carrier may take with it when a programme moves, which is the question that actually arises.

Part two: rating models and the filing problem

The model is the competitive asset and the sector's most valuable unpatentable thing. Rating plans, segmentation variables, credit-based scores, catastrophe adjustments, and increasingly machine learning components determine profitability more than any wording does.

Patent protection is largely unavailable. A method of rating risk is an abstract idea implemented on a computer, and Alice Corp. v. CLS Bank International and Bilski v. Kappos dispose of most claims. Where a technical improvement exists — a specific data structure, a computational technique with a measurable performance effect — the Enfish, LLC v. Microsoft Corp. route is available, and it rarely covers the actuarial substance.

So the protection is trade secrecy under 18 U.S.C. § 1836 and 18 U.S.C. § 1839, with the reasonable measures analysis of Rockwell Graphic Systems, Inc. v. DEV Industries, Inc..

And the filing regime cuts against it. Rate and form filings require actuarial support, and the filed material may be public depending on the state and the confidentiality treatment available. A carrier must therefore know, before designing the model, which elements will have to be filed, which may be submitted under a confidentiality claim, and whether that claim will be honoured.

Confidentiality claims must be made properly and at the time. Material submitted without a claim is generally public, and a claim asserted after a request has been made is usually too late. Mark, claim, and justify at submission.

Public records requests are a live channel. Competitors and journalists request filings. A carrier that assumed obscurity rather than confidentiality will find its segmentation approach described in a trade publication.

Regulatory scrutiny of models is intensifying, with attention to unfair discrimination, proxy variables, and the use of external data in underwriting and pricing. That scrutiny requires explanation, which requires documentation, which increases the disclosure surface.

Model governance discipline — inventory, ownership, validation, monitoring, and documentation — serves both the regulatory obligation and the trade secret position, because a model nobody has documented cannot be shown to have been treated as secret either.

Vendor models inherit all of this. A carrier using a third-party score must satisfy its own filing and explanation obligations using material the vendor will not disclose. Negotiate a validation and filing support package at procurement, including the vendor's commitment to support a confidentiality claim and, if necessary, to deal with the regulator directly. See Buying a Model and the AI Procurement Checklist.

Part three: data sources and their licences

Underwriting and pricing run on purchased data, and the licences constrain the product more than the actuarial science does.

Consumer credit and report data carries statutory obligations on permissible purpose, accuracy, dispute handling, and adverse action notices. The licence sits on top of the statute and typically restricts use to specified lines and purposes.

Motor vehicle records are subject to statutory access restrictions with defined permissible uses.

Medical and health information carries its own regime where a covered relationship exists, and falls into the general framework otherwise. See The App That Knows Your Diagnosis.

Property characteristic data, aerial and satellite imagery, and geospatial layers are licensed with scope, derivative work, and redistribution restrictions that frequently prohibit exactly what a carrier wants to do — retain a derived attribute after the licence ends.

Catastrophe model outputs are licensed with restrictions on disclosure of results, on use for purposes other than the licensed line, and on reporting to reinsurers or regulators without permission.

Telematics and connected device data raises consent, sensitive category, and secondary use questions, plus the question of who owns the derived driving score. Where a device is supplied by a vendor, the vendor's terms may claim the aggregate.

Aggregation and benchmarking across carriers is the industry's recurring structure and its recurring competition question, since pooled loss data assembled by a common vendor can shade into information exchange.

The recurring drafting problems are: derived data definitions that leave a carrier unable to keep a computed attribute; termination terms that require deletion of data embedded in models; audit rights exercised on a multi-year lookback; and affiliate use that a group structure quietly exceeds. See Selling Something You Cannot Own and the Data Licensing Checklist.

Part four: distribution, delegated authority, and the ownership gap

The place where insurance intellectual property most often goes wrong is a contract about something else.

Managing general agents and programme administrators underwrite on a carrier's paper, develop the wording, build the distribution, and hold the customer relationship. The agreement specifies authority limits, commission, and reporting, and is frequently silent on who owns the wording, the rating plan, the submission data, the loss data, and the renewal rights.

On termination, everything is contested. Can the administrator take the programme to another carrier? Can the carrier continue writing the book without the administrator? Who owns the manuscript wording? Who owns the model built on the book's loss experience? Who may contact the policyholders?

Ownership of expirations is the traditional formulation and it is not the same question as ownership of intellectual property. A broker may own expirations while the carrier owns the wording, or the reverse, and the two provisions frequently contradict each other.

Broker agreements raise the same questions at a smaller scale, with the additional feature that the broker holds the client relationship and regards the submission data as its own.

Wholesale and binding authority arrangements add another layer, with the wholesaler holding wordings and systems that neither the retail broker nor the carrier controls.

Reinsurance and fronting arrangements distribute the economics without distributing the intellectual property, and a fronting carrier may find it has lent its paper to a programme whose wording it does not own.

Practice notes. Address ownership expressly in every delegated agreement, distinguishing base wording, endorsements, rating plan, submission data, loss data, models, and customer records. Provide for each on termination, with a licence-back where the commercial reality requires shared use. Address non-solicitation and renewal rights separately from ownership. And record who drafted what, because in five years nobody will remember.

See the Channel Partner IP Checklist for the analogous structure in other distribution relationships.

Part five: claims systems, vendors, and the technology layer

Core systems are licensed and long-lived. Policy administration, claims, and billing platforms are deployed for decades, heavily configured, and integrated into everything. The intellectual property terms determine whether the carrier can ever migrate.

Configuration and customisation ownership is the recurring gap. A carrier that spends years configuring a platform has created something valuable, and the vendor's standard terms frequently claim it or restrict its use elsewhere.

Data portability determines whether migration is possible. Export format, completeness, historical depth, and the fate of unstructured claim notes should be settled at signature.

Escrow matters for the same reason it matters in regulated utility procurement: a thirty-year asset and a vendor who may not survive it. Escrow with verification, releasable on abandonment as well as insolvency.

Claims data is the carrier's most sensitive holding — medical records, investigation material, reserve information, and privileged communications. Vendor access terms should reflect that, and the privilege position on investigation material should be established deliberately. See Protecting Privilege in an IP Matter.

Insurtech partnerships invert the usual posture: a small vendor with a novel product, a large carrier with the distribution and the capital, and a partnership agreement that must allocate improvements, jointly developed material, and exclusivity. Address background and foreground intellectual property expressly, as in any collaboration. See Whose Invention Is It?.

Open source in core and peripheral systems requires the ordinary inventory and obligation analysis. See Copyleft and Consequences.

And security incidents touch claims data, underwriting data, and model assets simultaneously, with a notification matrix spanning breach statutes, regulators, reinsurers, and contractual clocks. See Running a Data Breach Response and the Cybersecurity Governance and Disclosure Toolkit.

Part six: brand, product names, and the filing layer

Product names require both trademark clearance and regulatory approval. A product name appears on a filed form, and a name that clears the register may still be refused or may require refiling if changed.

Descriptiveness is the chronic obstacle, since insurance product names describe what they cover. Registration runs under 15 U.S.C. § 1051 with the bars in 15 U.S.C. § 1052, and acquired distinctiveness is the usual route for a descriptive product name that has been used for years.

Carrier group naming raises statutory restrictions on terms implying particular status, plus the ordinary confusion analysis under 15 U.S.C. § 1114 and 15 U.S.C. § 1125.

Distribution marks — agency branding, co-branded programmes, affinity arrangements — require licence terms with quality control, because a mark licensed without supervision risks abandonment. See Naked Licensing.

Advertising review obligations apply to marketing material, with state-specific requirements and record retention.

And the approval sequence should run in parallel, not in series: trademark search, regulatory naming check, form filing, and launch, with the search conducted before the actuarial work is committed to a name. See the Financial Services Branding Checklist and the Brand Name Approval Toolkit.

Clause bank

Wording ownership in a delegated authority agreement. (a) The Base Wordings listed in Schedule A are and remain the property of Carrier. (b) Copyright in each Manuscript Endorsement developed under this Agreement vests in [Carrier / Administrator] on creation, and [the other party] hereby irrevocably assigns all right, title and interest in it. (c) On termination, [Administrator] may continue to use the Manuscript Endorsements identified in Schedule B on a non-exclusive, royalty-free basis solely for the Programme, and shall cease all other use. (d) Neither party may register copyright in a Manuscript Endorsement without notifying the other. (e) Each party shall maintain a record identifying the author of each Manuscript Endorsement it drafts.

Rate filing confidentiality support. Vendor shall, at Carrier's request and at [no additional charge / agreed rates]: (a) provide actuarial documentation sufficient to support Carrier's rate filing in each state in which Carrier writes the Programme; (b) support Carrier's claim of confidential treatment for that documentation, including by providing the justification required by the relevant state; (c) where a regulator declines confidential treatment, consult with Carrier before any disclosure and, at Carrier's election, deal with the regulator directly under a protective arrangement; and (d) notify Carrier of any public records request received by Vendor relating to the Programme within [5] business days.

Data licence derived attribute retention. Licensee may compute Derived Attributes from Licensed Data and may retain and use Derived Attributes after termination, provided that: (a) the Derived Attribute does not permit reconstruction of any Licensed Data record; (b) the Derived Attribute is computed at the [policy / risk] level and not stored as a copy of Licensed Data; and (c) Licensee ceases to refresh any Derived Attribute after termination. For the avoidance of doubt, models trained on Licensed Data may remain in production for [period] after termination, after which Licensee shall retrain or retire them.

Programme termination and transfer. On termination of this Agreement: (a) Ownership of the items in Schedule C is allocated as stated in that Schedule; (b) Carrier shall have a non-exclusive, perpetual, royalty-free licence to use the Rating Plan solely for the run-off of policies written under the Programme; (c) Administrator shall deliver the Submission Data and Loss Data in the format specified in Schedule D within [30] days; (d) neither party shall solicit the Programme's policyholders for [period], save that renewal offers by Carrier on existing policies are permitted; and (e) each party shall certify deletion of the other's Confidential Information within [60] days, excluding regulatory retention copies.

Core system migration assistance. On expiry or termination, Vendor shall provide Migration Assistance for a period of [12] months at the rates in Schedule E, comprising: complete export of all Customer Data including unstructured claim notes and document attachments in [specified format]; the data dictionary and entity relationship documentation; reasonable cooperation with Customer's successor supplier; and continued operation of the Services during the migration period on the same terms. Vendor shall not condition Migration Assistance on the settlement of any disputed amount.

Failures that recur

A manuscript wording drafted by a broker and used by a carrier for a decade, with no assignment, discovered when the programme moves.

Bureau form use continuing after the subscription ends.

A rating model filed without a confidentiality claim, then requested and published.

A confidentiality claim asserted after the request arrives, too late.

A derived attribute embedded in a model that the data licence requires to be deleted on termination.

Affiliate use of a licensed dataset across a group structure with a single-entity licence.

A delegated authority agreement silent on ownership, litigated at termination.

Ownership of expirations confused with ownership of intellectual property, producing contradictory provisions in the same document.

A vendor model deployed without filing support, leaving the carrier unable to justify its rates.

Core system configuration owned by the vendor, discovered at migration.

A product name committed to before the trademark search, requiring a refiling.

And a co-branded distribution mark licensed without quality control, weakening the mark it was meant to extend.

Worked scenarios

A programme that moves. A managing general agent has run a specialty programme on a carrier's paper for eight years. It drafted the manuscript wording, built the distribution, and holds the loss data. The relationship sours and the agent moves the programme to a competitor. The carrier asserts ownership of the wording; the agent asserts ownership of expirations and of the wording it wrote. The agreement addresses commission and underwriting authority in twelve pages and ownership in none. The dispute settles on a split — the carrier keeps the base wording, the agent takes the endorsements under a licence-back, the loss data is shared, and neither solicits for a year — at a cost that exceeds by an order of magnitude the price of the schedule that should have been attached at the outset.

A model in a public filing. A carrier develops a segmentation approach that materially improves loss ratio in a competitive line. It files the rate change with full actuarial support and no confidentiality claim, on the assumption that filings are obscure. A competitor's pricing team requests the filing, reads the segmentation approach, and replicates it within two quarters. Nothing unlawful occurred. The carrier's mistake was procedural: the confidentiality claim was available, had to be asserted at submission with a justification, and was not. The remediation is prospective — a filing protocol that requires a confidentiality assessment before every submission, with a documented decision either way.

A derived attribute stranded by termination. A carrier licenses a property characteristics dataset and computes a composite risk attribute from it. The attribute is stored on every policy record and drives the rating plan. The licence terminates. The licensor's position is that the attribute is derived data requiring deletion; the carrier's position is that a computed score is its own work. Neither position is obviously right, because the derived data definition was drafted for a different use case. The carrier's exposure is that it cannot rate its own book. The clause that would have prevented it is two sentences and is in the clause bank above.

An insurtech partnership with no foreground allocation. A carrier and a small technology company build a product together. The technology company brings a claims triage engine; the carrier brings underwriting expertise and data. Improvements are made jointly over two years. The agreement says the parties will "discuss in good faith" the ownership of jointly developed material. The partnership ends. Neither party can use the improved engine without the other, both want to, and the good faith discussion produces litigation. Foreground and background allocation is a schedule, not a sentiment.

A ninety-day programme

Weeks 1–2. Inventory the wordings. Every base form, its source, and its licence. Every manuscript endorsement, its author, and the instrument that assigned it. Flag every endorsement with no assignment.

Weeks 3–4. Inventory the delegated arrangements. Every managing general agent, programme administrator, broker, and wholesaler agreement, with the ownership provisions extracted and the gaps recorded. Rank by premium volume.

Weeks 5–6. Inventory the models. Every rating and underwriting model, its owner, its inputs, its filing history, its confidentiality treatment, and whether it is internal or vendor-supplied. Flag every model filed without a confidentiality claim.

Weeks 7–8. Inventory the data licences. Every source, every permitted use, every derived data definition, every affiliate and termination provision. Compare against actual use and quantify the gaps.

Weeks 9–10. Review the core system and vendor agreements for configuration ownership, portability, escrow, and migration assistance. Rank by migration difficulty rather than by contract value.

Weeks 11–12. Review the product and distribution marks: registration status, descriptiveness exposure, quality control in co-branded arrangements, and the parallel clearance sequence for new products.

Throughout. Amend the delegated agreement template to include an ownership schedule, and use it on every renewal from week one. Renewals are the only moment these terms improve without being paid for.

Two documents worth keeping current

The wording register. One row per wording: name, base form and its source, endorsements, author of each, assignment instrument and date, programmes it is used in, filing status by state, and registration status. This answers the question that arises on every programme transfer and on every acquisition, and it takes an afternoon a year to maintain.

The model register. One row per model: name, owner, business use, inputs and their sources, vendor if third-party, filing history by state, confidentiality claim status and outcome, validation date, monitoring cadence, and documentation location. It serves the regulator, the actuary, the trade secret position, and the acquirer simultaneously.

Part seven: the wording as an asset in a transaction

Insurance transactions — renewal rights deals, portfolio transfers, agency acquisitions, and carrier mergers — turn on intellectual property questions that the deal documents frequently treat as afterthoughts.

A renewal rights transaction transfers the right to solicit renewals of a book without transferring the liabilities. What it does not automatically transfer is the wording, the rating plan, the underwriting guidelines, the submission data, or the systems configuration. A buyer that acquires renewal rights and discovers it cannot use the wording has bought a list.

An agency or brokerage acquisition transfers the client relationships and, if properly documented, the expirations. It should also address the agency's own manuscript wordings, its proprietary programme structures, its data, and the marks under which programmes are distributed.

A portfolio transfer or loss portfolio transaction moves liabilities and reserves, and should address which party may continue to use the wording for run-off and for new business.

A carrier merger inherits every bureau subscription, every data licence, every delegated agreement, and every unresolved ownership question in each. The diligence should sample rather than schedule: pick five programmes and trace the wording, the model, the data, and the distribution ownership through to a documented answer.

Change of control provisions in bureau subscriptions, data licences, and core system agreements can require consent, trigger repricing, or permit termination. Identify them early, because a consent nobody sought becomes a closing condition nobody planned for.

And the run-off question is the one most often missed. After a transaction, somebody must service policies written under a wording and rated by a plan that the transaction may have moved to the other party. A run-off licence — non-exclusive, perpetual, limited to the existing book — costs a sentence and prevents a genuine operational failure.

Part eight: competition considerations

Insurance intellectual property practice runs unusually close to competition law, and the sector's habits make the proximity easy to forget.

Shared statistical plans and pooled loss data are long-standing industry structures, produced by advisory organisations under regulatory oversight. They function because they operate at an aggregate level and under a specific legal framework, and a private arrangement replicating them without that framework is a different proposition.

Vendor-mediated benchmarking is the modern version. A technology vendor serving many carriers can produce comparative pricing and loss analytics of great value, and the closer those outputs get to current, identifiable, forward-looking pricing information, the more they resemble an exchange between competitors.

Design the aggregation deliberately: minimum contributor counts, no attribution to any contributor, historical rather than forward-looking data, sufficient aggregation and lag, and a documented methodology. These are the same controls described in the Competitive Intelligence and Benchmarking Toolkit.

Joint development between carriers — a shared wording, a common platform, a jointly funded model — needs a competition assessment alongside the intellectual property allocation, and needs information barriers between the participants' commercial teams.

Delegated authority arrangements can raise resale price and territorial questions where they constrain the agent's dealings with other carriers, and exclusivity provisions should be scoped with that in mind.

And the practical advice is to route any arrangement that pools information between competitors through a competition review before the intellectual property terms are drafted, because the answer may change the structure entirely rather than merely the wording.

Part nine: telematics, connected devices, and the new data

The sector's newest data sources raise questions the older ones did not, and the contracts have not caught up.

Motor telematics produces continuous driving behaviour data — speed, braking, cornering, phone use, time of day, and location. It is behavioural, it is granular, and in several state regimes location data is a sensitive category requiring specific treatment.

The device layer matters. Where the data comes from a vehicle manufacturer, the manufacturer's terms govern; where it comes from a dedicated device, the device vendor's terms do; where it comes from a mobile application, the platform's terms and the application's own disclosures do. Three routes, three sets of constraints, and a product built on one that migrates to another inherits new terms.

Connected home and commercial property sensors — water leak detection, temperature, occupancy, fire — raise the same questions with an added consent problem, because the sensor observes people who are not the policyholder.

Wearables in life and health underwriting engage health information rules, genetic information prohibitions where family history is implicated, and the general sensitivity of physiological data. The analysis is closely parallel to the athlete data problem described in Every Step Recorded and the Athlete Data Checklist.

Aerial and satellite imagery used for property underwriting is licensed with restrictions on derivative products and redistribution, and the derived attribute problem described above applies directly.

Claims-side sensor data — the vehicle's own crash record, the building system's log — is evidence, and its handling should reflect that: preservation, chain of custody, and an assessment of who else has a right to it.

Consent architecture is the recurring weakness. A policyholder who agreed to telematics for a discount has consented to something; whether that something includes retention after cancellation, use in claims disputes, sharing with an affiliate, or contribution to an industry dataset depends entirely on what the disclosure said, and most disclosures were drafted for a simpler product.

And the discount is the leverage question. A programme in which participation is technically optional but economically compulsory raises the same consent-quality problem that arises wherever a measured party has no realistic alternative. The honest test is whether the arrangement would survive being described plainly to the policyholder — which is a better guide to regulatory durability than any current rule.

Part ten: enforcement, and when it is worth it

The sector's default is not to enforce, and that default is usually right. It is worth knowing when it is not.

Verbatim copying of a distinctive manuscript wording by a direct competitor, where the wording was registered before the copying and where the copied language is expressive rather than dictated by the concept, is the one fact pattern that supports a claim worth bringing. Registration timing controls the remedy under 17 U.S.C. § 412, and Fourth Estate Public Benefit Corp. v. Wall-Street.com, LLC controls when suit may begin.

Departure of an underwriting or actuarial team with model documentation, rating plans, or submission data is a trade secret matter under 18 U.S.C. § 1836, and the forensic timeline usually decides it. Preserve devices before the standard deprovisioning schedule runs.

A managing general agent taking the programme is a contract matter first and an intellectual property matter second, and the leverage comes from the ownership schedule rather than from a copyright claim.

A vendor exceeding its licence — using a carrier's loss data to build a product sold to competitors — is a contract claim with a trade secret overlay, and it is worth pursuing because tolerating it establishes a practice.

Brand disputes follow ordinary trademark practice under 15 U.S.C. § 1114 and 15 U.S.C. § 1125, and are more common in distribution than in product naming.

What is usually not worth it: pursuing a competitor for adopting a market-standard clause; asserting copyright in language that is effectively dictated by the coverage concept; and litigating a data licence interpretation with a licensor whose product the business still needs.

And the discipline that makes enforcement possible is unglamorous: register the wordings that matter, record who drafted what, mark and access-control the models, and keep the ownership schedules current. A carrier that has done those four things has options; one that has not is negotiating from a description of what it believes it owns.

One paragraph to remember

In insurance the product is a document somebody else may own, the pricing is a secret the regulator requires you to explain, the data is licensed on terms that decide what the product can be, and the distribution agreement that governs all three was drafted to address commission. Attach an ownership schedule to every delegated arrangement, claim confidentiality at the moment of filing rather than when the request arrives, negotiate the derived attribute definition before the attribute is embedded in a rating plan, and record who drafted each endorsement while somebody still remembers.

The five-document test

The quickest diagnostic on an insurance intellectual property position takes an afternoon. Choose one programme and ask for five documents: the base wording with its licence or source; the assignment for each manuscript endorsement; the ownership schedule from the delegated authority agreement; the confidentiality claim filed with the rate submission; and the data licence provision permitting each external attribute used in the rating plan.

A programme that produces all five is unusually well run. A programme that produces two is the ordinary case. A programme that produces none is not unusual either, and its owner will discover the position at the worst possible moment — when the agent leaves, when the competitor files a records request, or when the acquirer's counsel asks the same five questions with a purchase price attached.

Key Authorities at a Glance

Copyright. 17 U.S.C. § 102 on subject matter; 17 U.S.C. § 101 on work made for hire; 17 U.S.C. § 201 on initial ownership; 17 U.S.C. § 204 on transfers; 17 U.S.C. § 103 on compilations; 17 U.S.C. § 106 on exclusive rights; 17 U.S.C. § 107 on fair use; 17 U.S.C. § 412 and 17 U.S.C. § 504 on remedies; Feist Publications, Inc. v. Rural Telephone Service Co.; Baker v. Selden; Fourth Estate Public Benefit Corp. v. Wall-Street.com, LLC.

Patent and eligibility. 35 U.S.C. § 101 with Alice Corp. v. CLS Bank International, Bilski v. Kappos, and Enfish, LLC v. Microsoft Corp.; 35 U.S.C. § 112.

Trade secret. 18 U.S.C. § 1836; 18 U.S.C. § 1839; Rockwell Graphic Systems, Inc. v. DEV Industries, Inc.; PepsiCo, Inc. v. Redmond.

Trademark. 15 U.S.C. § 1051; 15 U.S.C. § 1052; 15 U.S.C. § 1114; 15 U.S.C. § 1125; 15 U.S.C. § 1127 on abandonment; Barcamerica International USA Trust v. Tyfield Importers, Inc. on naked licensing.

Data and access. 18 U.S.C. § 1030 with Van Buren v. United States; unfair and deceptive practices under 15 U.S.C. § 45. For the statutory consumer data regimes that govern underwriting inputs, see Fair Credit Reporting and Driver's Privacy Protection.

Preemption and state law. 17 U.S.C. § 301 governs whether a state law claim over copying a wording survives; see What the Copyright Act Kills.

| Authority | Governs | Practical consequence | | --- | --- | --- | | 17 U.S.C. § 102 | Wordings as literary works | Copyright exists but is thin | | Feist | Originality | Modest creativity suffices | | Baker v. Selden | Merger | Functional language unprotected | | 17 U.S.C. § 204 | Transfers | No writing, no assignment | | 17 U.S.C. § 412 | Remedies | Register before infringement | | Alice | Eligibility | Rating methods are not patentable | | 18 U.S.C. § 1836 | Trade secrets | Where models are protected | | Rockwell | Reasonable measures | Documentation serves both purposes | | Rate filing regimes | Disclosure | Claim confidentiality at submission | | 15 U.S.C. § 1052 | Registrability | Descriptiveness bars most product names | | Barcamerica | Naked licensing | Supervise co-branded distribution | | 17 U.S.C. § 301 | Preemption | Limits state claims over copying |