It may sound counterintuitive, but you don’t want to completely “block AI”.
Instead you want to retain monetisation power: membership, subscriptions, syndication, licensing, B2B research, or some mix that doesn’t collapse the moment reader discovery changes again in future.
The third instalment of a three-part series on how IP is the best tool in your arsenal. To be blunt: IP is your leverage, but it’s not a magic shield. It only becomes leverage when it supports a revenue mechanism and is paired with trust, attention, and first-party channels/data.
If you read Parts 1 and 2 and still feel stuck, that’s normal. Most advice on AI and publishing fails for a simple reason: it’s mostly all hype, and no one talks about the tactics that actually work.
I often talk about how incentives drive outcomes. This article is about how we create new incentives, so everyone wins.
What follows is designed for various tiers of the independent publishing sector:
small-team operations of 2-5, larger teams of up to 20, and individual freelancers producing long-form work (features, investigations, archives, interviews, research), not daily news churn.
I. Reframe the context: rights don’t earn revenue on their own
There are three things we can say with reasonable confidence from the evidence we’ve covered in parts 1 and 2:
Direct audience payment for distinctive, trusted work is more resilient than ad-only models, especially when generic summarisation obliterates discovery.
Europe is directionally pushing AI providers (albeit slowly) toward copyright compliance and training-data transparency, which increases the value of rights-managed archives.
Litigation and enforcement are structurally asymmetric; smaller players cannot build a business model around lawsuits. That’s why I emphasise simple tactics that leverage your size and position:
rights clarity, machine-readable signals, membership, first-party channels, and collective structures.
So the operational question changes from this nebulous idea of “how do I win the AI war?”, to the more precise: how do I make my work harder to free-ride, easier to monetise, and less dependent on any single distribution gatekeeper?
II. The four archetypes (defined by revenue mechanics + rights posture)
This is not an exhaustive list, simply the groups of independent publishers and journalist who I can help right now. Many face similar pains and problems, and have similar goals, yet each group has different different weak points, with particular nuances.
One of the murkier challenges facing everyone is managing rights and relationships with AI firms who want to use your work.
I’ve written before about the symbiotic, co-dependent relationship between publishers, journalists, and AI companies. Each party needs the other, for different reasons. Journalists and publishers need the AI companies to replace the lost traffic in an era of zero-click search, and the AI firms need libraries of “high-fidelity” data, ie. the articles in your catalogues that are subject to a highly rigorous editorial process.
A) The Licensed-by-others Operator (freelance / solo)
You earn via third-party commissions, retainers, publishing platform deals (e.g. Substack), or specific contracts where someone else controls distribution.
The risks are:
your rights are signed away, reuse terms are vague (usually not to your benefit), and your archive becomes someone else’s asset.
We’ve explicitly flagged rights-ownership and contract terms as a first-order factor for increasing your leverage.
B) The Audience-moat Publisher (membership/subscription/community-led)
Your revenue is direct reader payment. Your risk is substitution and conversion: if answers appear elsewhere and users don’t reach your paywall or newsletter sign-up, your funnel shrinks. You don’t need me to tell you that attention and trust are scarce, whereas generic information is not.
Therefore your moat must be relationship, context, participation, and clear terms.
C) The Small Magazine with a back catalogue
Your revenue is mixed (e.g. ads/subs/syndication/grants). Your risk is operational decay: contributor agreements vary, ownership and reuse is unclear, and the archive cannot be simply packaged or licensed cleanly.
Focus on rights consolidation and rights-cleared archives as these are economically valuable, but only if the rights are precise and clear.
D) The Niche research / data-heavy vertical
You monetise via B2B reports, briefings, training, consulting, and specialist products. Your risk is that the “collection” is valuable but unstructured: permissions are unclear, provenance is messy, and delivery is scattered across tools.
You can create leverage through database/collection rights (UK/EU) and the value of structured, hard-to-replicate datasets.
III. Baseline moves (for everyone, no additional resources required)
These are simple, evidence-based actions; useful whether LLM training ends up broadly lawful or increasingly licensable.
1) Rights inventory in one sitting
Make a single table: what exists (articles, photos, audio, datasets), who created it, where it’s stored, what agreement governs it, and what’s missing. We put “audit and consolidate rights” first for a reason: without this, you can’t license, syndicate, or enforce with confidence.
2) Contract triage: fix forward before you fix the past
Update templates for new work first (you stop the bleeding), then backfill the archive as capacity allows. To be explicitly clear, allocate rights for each of: use/re-use contracts, freelance agreements, and platform terms; retaining or reserving various use-cases increases your ability to license, repackage, and join collective schemes later.
3) Make your position legible + enforceable to machines
Use clear copyright notices and machine-readable directives (robots.txt; “noAI/noTrain” style signals where relevant). For EU-facing sites, consider explicitly reserving rights under DSM Directive Article 4(3) if you want to opt out of commercial TDM. This is a practical step that strengthens your position in scenarios that require opt-outs be respected.
Although it should be noted that robots.txt is becoming increasingly ineffective (there are several companies looking to address this problem, Writers’ Bloc being one of them - contact me directly if you want help here; see below).
4) Access + packaging hygiene
To reiterate, we discourage a “block everything” approach. I argue that creating real boundaries delivers better results: paywall integrity, sensible friction against scraping where appropriate, and, most importantly, packaging your archive into defined units aligned to your business model (collections, dossiers, annotated investigations, indexes). I cannot stress enough that rights-cleared, curated corpora and structured archives are the assets that are valuable, and become negotiable.
Brooke Hartley Moy of Infactory puts it well, as she argues that publishers need to think more like data companies. Your library is a treasure trove for the AI firms, so it’s in your interests to set it up in such a way that you get paid when they access it, and use the work therein.
5) Provenance record-keeping (boring, but decisive)
Can you show what you published, when, and under what terms? Evidence and traceability are prerequisites for enforcement and negotiation, especially in cross-border settings where arbitrage opportunities exist.
6) A simple stress test
If you do nothing else, run this one scenario: “If AI chatbots eat 30% of your search traffic, what breaks?” Ad-only models snap faster than subscriber/member-focused models; the point is to see which revenue lever needs strengthening first.
IV. The playbooks (ROI-ordered, start here)
A) Freelancer / Solo-operator (licensed by others)
Replace “all rights” defaults with reserved uses (syndication, archive reuse, AI reuse limits where negotiable).
Build a “rights pack” for your best long-form: what it is, proof of authorship, permitted uses, pricing logic.
Consolidate a direct channel (newsletter/site) so your archive is not trapped inside other people’s paywalls and terms.
Register high-value works where it improves remedies (US-specific, especially for major investigations).
Add contractual prohibitions on voice/likeness cloning where you have leverage.
B) Audience-moat Publisher
Make the “social contract” explicit: what members fund and what they get that generic summaries can’t replicate.
Turn archives into member benefits (special collections, annotated investigations, Q&As).
Own distribution: newsletters/apps/community, because zero-click products increase the value of first-party channels.
Track basic first-party data (cohort retention, conversion paths) to prove value and guide packaging decisions.
Use AI internally for drafts/analysis, but keep human authorship and fact-checking explicit to preserve trust and protectability.
C) Small newsroom with a back catalogue
Standardise contributor terms going forward; backfill top-value archive contracts next.
Build an internal index (“knowledge spine”) so you can package investigations/datasets without reinventing the wheel.
Create one syndication/licensing SKU (a defined archive package) and one member SKU (archive access + participation).
Publish a clear AI/IP policy to protect trust and set expectations.
Join alliances/collective structures: this is how small players reduce enforcement costs and gain negotiating scale.
D) Niche research / data-heavy vertical
Turn scattered materials into a structured collection (taxonomy + versioning).
Clarify permissions and provenance for the dataset/collection (especially relevant under UK/EU database right logic).
Productise outputs into repeatable briefings/reports/training—sell judgement and verification, not raw text.
Invest in basic analytics to show value to buyers and to focus on the highest-leverage beats.
Keep AI as a tool for analysis and synthesis, not as a substitute for the core asset: verified, context-rich reporting.
V. Build optionality that works in your favour
It’s painful to read, but our most defensible stance is also the least comforting: outcomes are jurisdiction-specific, enforcement is unequal, and “market dilution” theories remain contested.
So don’t bet your business on courtroom outcomes. Instead build a business that survives both scenarios, whether training stays broadly permissible and/or can capitalise if licensing expands:
clean rights, machine-readable signals, packaged archives, first-party channels, and a membership/B2B proposition rooted in trust.
If this lands with you, feel free to contact me directly, and I will help you with whatever you need and clarify anything if you get stuck.
Message directly on LinkedIn, send me an email ([email protected]), or book a short call, and I will personally help you. I want to help my most active subscribers.
Next week we explore why everyone thinks “data is the new oil”, and how publishers are the landowners sitting atop the oil fields.
Appendix:
Practical guardrails
Disclaimer:
I am not a lawyer and this is not legal advice. This is general information for educational purposes only and may not reflect the current law in your jurisdiction. Speak to qualified counsel before relying on any sample language.
A) Contract language (examples)
1) Ban AI training / datasets
The Client/Publisher shall not use, and shall not permit any third party to use, the Work (in whole or in part) to train, fine-tune, evaluate, or develop any AI/ML model, nor include it in any dataset for such purposes, without the Author’s prior written consent.
2) No onward transfer to AI firms / data brokers
The Client/Publisher shall not sublicense, sell, transfer, or provide access to the Work (or any compilation including it) to any AI developer, model provider, data broker, or dataset vendor without the Author’s prior written consent.
3) Survival + breach
These restrictions survive termination. Unauthorised AI/ML use is a material breach.
B) Website signals (examples)
WordPress: robots.txt (best single control point)
Add/edit: yourdomain.com/robots.txt
User-agent: *
Disallow: /wp-admin/
Disallow: /wp-includes/
# add your paywall/archive paths if relevant:
# Disallow: /members/
# Disallow: /archive/WordPress: optional meta tag (sitewide, in <head>)
Use only if you understand indexing implications:
<meta name="robots" content="index, follow, noai, noimageai">C) Substack (unfortunate reality)
Substack generally doesn’t let you reliably control <head> tags or robots.txt. Your strongest lever is access control: keep premium work paid/subscriber-only, and publish excerpts for public posts.