Enforcement leverage lives where your counterparty’s revenue lives

Ericsson and smartphone maker Transsion signed a global patent cross-licence this week, ending a dispute that stretched across four continents and nine jurisdictions — and nearly a decade of stalled negotiation before that. What changed? Days earlier, a court in Casablanca issued Africa’s first-ever standard essential patent injunctions, targeting a market where Transsion holds a commanding share.

The durable lesson is not about telecoms. It is that patent enforcement works when it is aimed at the counterparty’s actual revenue, not at the venues lawyers find most comfortable. A licensing negotiation is a pricing conversation, and price follows leverage. If you hold rights, map where your counterparty makes money and ask whether you can credibly act there. If you are the one paying royalties, the emerging-market safe harbours you have been relying on are closing. Either way, structure matters as much in settlement as in suit — a point explored in The Reverse Payment You Didn’t Know You Made, on how the economics buried inside a patent settlement can carry risks of their own.

In the AI and copyright fight, records are the battlefield

In the consolidated US litigation over AI training on news content, major publishers asked the court to sanction OpenAI, alleging the company withheld datasets and output logs central to the case. Whatever the court decides, notice what the fight is actually about: not fair use doctrine, but records — what data was used, where it came from, and who can prove it.

That is the evergreen point for any organisation building or deploying AI. Data provenance is no longer an engineering detail; it is litigation evidence, licensing currency and due-diligence material in every future transaction. If you cannot say what went into your models — or what your teams are feeding into someone else’s — you are carrying unquantified risk. The same discipline problem arises inside businesses adopting AI agents, as discussed in Your AI Agent Won’t Keep a Secret: when systems generate and move information without a human in the loop, information governance becomes IP strategy.

A patent asserted is a patent audited

At Europe’s Unified Patent Court, BioNTech and Pfizer saw off an infringement claim over their COVID-19 vaccine when the Munich division not only rejected infringement but revoked the asserted mRNA patent across Germany, France and Sweden in a single stroke.

Two lessons travel well beyond pharma. First, for patentees: asserting a right invites a validity audit, and in the UPC a loss is a loss everywhere at once. Before you enforce, pressure-test the asset as ruthlessly as your opponent will. Second, for operating companies: the same central mechanism that threatens pan-European injunctions can deliver pan-European freedom to operate — a revocation counterclaim is a strategic weapon, not just a defence. The UPC’s expanding reach, and what it means for your patent litigation map, is examined in One Court, Many Borders: Why the UPC Just Redrew Your Litigation Map.

The connecting thread

Strip out the industries and one pattern remains: in each development, the decisive asset was infrastructure — a venue strategy matched to the counterparty’s markets, a defensible data trail, a patent that could survive hostile scrutiny. Rights on paper did not decide anything this week. Preparation did.

Three things to check this quarter: Where would enforcement (by you or against you) actually bite commercially — and does your IP portfolio management reflect that map? Could you document, today, what data your AI tools have consumed? And if you asserted your most valuable patent tomorrow, would it survive the audit? If any answer is uncomfortable, that is the work — and it is far cheaper done now, with your patent attorney or IP strategist, than mid-dispute.

That tension sits at the heart of a recent Harvard Business Review piece by Rita McGrath, The Power of Strategic Centering, which argues that in an economy where roughly 90% of corporate value now lives in intangible assets, the old anchors of defensible positions and durable physical assets no longer hold. Her central example is instructive: Novartis chose to center on innovative medicines, and a radioligand cancer therapy that looked like a gamble on a spreadsheet became an obvious move. The conglomerate was worth ~$190–200bn; the focused parts are now worth roughly twice that.

The lesson for IP-intensive organisations is that coherence, not breadth, is what compounds value — and intellectual property is where that coherence either becomes real or stays theoretical. (Read the article)

For in-house IP leads and GCs, here is the durable takeaway: your patent portfolio, trade secrets and brands are the operational expression of your company’s chosen centre — or evidence that it doesn’t have one. A business centred on a mission should be building IP around the problem, not a single technology, so its filings survive the next platform shift. A technology-centred company (think Fujifilm following its chemistry into cosmetics and medical imaging) should hold its capabilities as transferable assets and protect them accordingly, not lock them to a dying market.

The practical test is simple: pull your portfolio and ask whether each asset advances the centre or merely accumulated out of habit. Assets that no longer serve the strategy are cost and clutter; the gaps where the centre is exposed are your real risk. Aligning IP strategy with business goals isn’t a compliance exercise — it’s how you make the painkiller-versus-radioligand argument disappear before it starts. If you’re weighing a portfolio against a strategy that may have quietly drifted, that’s exactly the conversation worth having now, deliberately, rather than discovering the mismatch mid-transaction.

Two strategic signals matter for in-house IP counsel, GCs and founders weighing whether and how to fight. First, contrast the main appeal with the unjustified-threats claim, where the Court applied the orthodox rule that costs follow the event — a reminder that a clean, separable win is rewarded cleanly, while a sprawling multi-issue case invites discounting. The discipline, then, is to prune your grounds to the ones you can actually carry, because every issue you lose is an issue you may fund for the other side. Second, watch the enforcement-letter trap: the parties ended up litigating an unjustified-threats appeal because a demand letter was sent over a design whose ownership was contested. Under section 77 of the Designs Act 2003 (Cth), a threatened party can seek a declaration that the threats are unjustified, an injunction and damages — so a cease-and-desist letter you can’t back in court becomes its own liability, a pattern that echoes the entitlement risk in When a Departing Engineer Files a Patent and the disclosure discipline in Be prepared before you engage. The takeaway is unglamorous but durable: fix ownership before you build, narrow the fight before you file, and never send a threat you aren’t ready to prove.

The strategic lesson is broader than spirits, and it cuts against a common assumption that descriptive or geographic names are simply unregistrable. They are hard to register — but a place-of-origin name can become a protectable, premium-commanding asset if you build the evidentiary and regulatory groundwork early. Three things this case rewards: first, documented use over time — the win came from years of consistent labelling and import records, not from the elegance of the legal submissions, the same “build it before the fight” pattern visible in Built Before the Fight: What May’s IP Decisions Reward. Second, a credible rulebook: a certification mark is only as strong as the production and quality rules behind it, and those rules define who may use the name and on what terms. Third, a realistic view of the monopoly’s edge — the whole fight turned on whether other honest traders need the word, the same boundary between private right and public freedom explored in When the Monopoly Ends, Your Shape Has to Stand on Its Own. For founders and in-house counsel whose value is tied to origin — a regional food, a method, a heritage product — the takeaway is to treat the name as IP from day one: standardise how it appears on every unit, keep the sales and promotion records, and codify the rules before a competitor’s generic use hardens the term against you. Origin can be owned — but only by the business that proves it.

Precision: A Right Is Only as Strong as Its Weakest Document

Broad and strong are not the same thing. The Full Court’s best-method ruling on divisional strategy showed broadened claims slipping to a later priority date and falling to prior art — the same fragility that, in the US, cost Pfizer’s Paxlovid patent its priority date over a single character, one of three signals from a single week. Precision also decides who keeps an invention: the departing-engineer case was won on the paper trail — assignment clauses, dated records, contemporaneous emails. And in litigation, the discipline runs to your pleadings, where strategic amendments to invalidity grounds rewarded a timely, coherent theory and refused a diffuse one.

Stewardship: Rights You Don’t Work Quietly Lose Their Edge

A site-blocking injunction was extended past its tenth year only after the applicants pruned dead domains and re-certified that the targets were still infringing. Enforcement is an operating commitment, not a trophy. The same logic scales up: the pharmaceutical patent-cliff timetable — ~1,600 extended Australian patents — is effectively a public calendar of competitor exclusivity, engineered years ahead.

The Stewardship Gap Is Now an AI Problem

The burden is shifting fast. The rise of high-impact individual contributors means one person now generates a team’s worth of tacit know-how that walks out the door unless captured. Agentic AI sharpens the edge: your AI agent won’t keep a secret, disclosing confidential information autonomously and leaving discoverable records unlikely to attract privilege — so give it minimum data access, attach provenance, and log its sources. Underwriting all of it is attention at the top, and your IP strategy is only as good as your board’s questions.

What to Watch

The precision standard keeps rising, and the cheapest insurance — clean records, disciplined claims, diarised renewals — is sometimes neglected. The decisions ahead are operational: who monitors filings by former staff, who certifies an order still bites, who governs the AI agents acting on your behalf, and who at board level is asking. Treat IP as a standing discipline, or discover too late what decayed while no one was watching.

The IP consequences look familiar but behave differently. Copyright risk is amplified because the human signing off has less visibility of what the agent drew from. Confidential information can be disclosed autonomously, raising the prospect of breach of confidence without anyone deciding to share anything. And agents produce a large volume of text about their own actions — discoverable, and unlikely to be protected by legal professional privilege.

The strategic takeaway is that deploying an agent is an act of delegation, not a software rollout — and you don’t hand a delegate your trade secrets without limits. For IP-intensive organisations, three disciplines do real work: give agents minimum viable data access (the narrowest permissions needed, revoked when the task ends); attach provenance and a verification status to AI-generated content, so you know what’s been checked; and log inputs, sources and model versions so any output can be traced to its origin.

None of this is a compliance chore — it’s how you capture the upside while keeping the assets that make the business valuable intact.

Worth a read: https://www.governanceinstitute.com.au/advocacy/governance-in-the-age-of-agentic-ai/

Start with the clearest signal. Getty Images announced a multi-year display partnership with OpenAI, putting its licensed images inside ChatGPT — and its share price roughly doubled on the news. Read the structure, not just the headline. Getty, once the loudest litigant against AI image-scraping, carved out a display right and licensed it while keeping training rights separate. That is AI and copyright strategy in practice: granular rights, licensed deliberately, priced as an asset. The durable takeaway for any content owner is that a well-documented, rights-cleared catalogue is now a revenue line inside AI products. The mistake it helps you avoid is treating your library as a single undifferentiated blob — the value sits in your ability to slice rights and license them on your terms.

The second signal is a warning. In Enanta v. Pfizer, the Federal Circuit invalidated patent claims tied to Pfizer’s Paxlovid compound because a one-character gap between the provisional application and the issued claim cost the patent its priority date, letting intervening prior art anticipate it. For anyone who has ever rushed a provisional patent filing, the lesson is blunt: your priority date is only as strong as the literal disclosure in your earliest document. Sloppy drafting is not a clerical risk — it is a commercial one, capable of handing a rival a complete defence. This is exactly where a patent attorney earns their fee, and where in-house counsel should be auditing whether provisionals genuinely describe what the company later claims. Meanwhile, the US House Judiciary IP Subcommittee convened a hearing on whether decades-old copyright and patent frameworks still fit an AI-driven economy — a reminder that the rules underpinning all of this are themselves in motion.

The thread tying the three together is ownership discipline. Know precisely what you own, document it well enough to survive scrutiny, and license it deliberately rather than defensively. The organisations that treat IP strategy as a commercial function — not a compliance afterthought — are the ones converting rights into revenue while others are still litigating.

A recent Federal Court decision extending a long-running music industry injunction for a further three years — taking it past the ten-year mark — is a useful reminder of how these orders actually earn their keep. Under section 115A of the Copyright Act, a copyright owner can obtain a carriage service provider injunction requiring ISPs like Telstra, Optus and TPG to disable access to overseas piracy sites. The strategic point sits in the housekeeping: the Court extended the order, but only after pruning dead domain names, swapping out respondents that were no longer carriage service providers, adding new corporate entities, and — notably — requiring the applicants’ solicitors to certify a good-faith belief that the targeted sites were still infringing. The order adapts because the people enforcing it keep feeding the Court accurate facts.

The durable lesson for IP-intensive businesses is that enforcement is an operational commitment, not a one-off win. Pirate operators register fresh domains the moment the old ones are blocked, so the value of a site-blocking order lies in the mechanism that lets you add new pathways by agreement with the ISPs, without running back to court each time — and in the discipline of actually using it.

Three things are worth building into your online piracy enforcement program: a monitoring routine that catches mirror and proxy domains as they appear; a documented, regular review so you can certify in good faith that targets are still live and infringing (and promptly release blocks on those that aren’t); and a diarised renewal well ahead of expiry, because these orders lapse if you let them. The broader principle is one we return to often — a right you don’t actively manage quietly decays. The competitive edge goes to the owner who treats enforcement as a standing process, with the evidence trail to back it.

For related reading, see Repeated infringement and failing to show up gets you 5x damages on how a defendant’s conduct shapes the remedy, IP update: vaccines, counterfeits, textbook piracy and tm translations on the wider piracy and counterfeiting enforcement landscape, and Fast and intuitive IP decision making on making sound enforcement calls under pressure.

Read the full judgment — Universal Music Australia Pty Limited v TPG Internet Pty Ltd (No 2) [2026] FCA 731.

Lenehan J’s analysis is a useful reminder of how the notional consumer’s imperfect recollection actually works: the earlier mark sits wholly inside the later one, the first two syllables are identical, and both words gesture toward the same idea — a cyclone. Visual similarity, aural similarity, and shared connotation combined to create a real, tangible risk of confusion, even though “CYCLON” reads as invented and “CYCLONIC” is a known English word. The Court was unpersuaded that consumers would meaningfully distinguish between them at the point of sale, particularly in fast-moving online and physical apparel channels where brand names sit side by side.

For IP‑intensive businesses, the strategic signals are worth pulling out.

First, clearance work needs to test marks against imperfect recollection — not side-by-side comparison — and weight the front of the word heavily, because that is what consumers carry away.

Second, adding a descriptive or grammatical suffix to an existing mark is a fragile differentiation strategy; if the root carries a recognisable idea, the suffix rarely rescues it.

Third, the Class 25 / Class 40 finding on “closely related” goods and services is a quiet but important point: recycling and textile-processing services are now commonly bundled with apparel retail, which expands the conflict zone for fashion brands well beyond their immediate product class.

And finally, the costs outcome — a $125,000 lump sum against a respondent who neither withdrew its application nor filed a submitting notice — underlines that disengagement is not a costless strategy once an appeal is on foot. Disciplined portfolio review, deliberate suffix and family-mark choices, and a clear exit plan when an opposition turns into litigation are all cheaper than the alternative.

Full judgment: https://www.judgments.fedcourt.gov.au/judgments/Judgments/fca/single/2026/2026fca0647

This is one of the more under-managed risks in any R&D-heavy business — and a recent Federal Court decision shows just how fast it can move once it surfaces. The Court ordered the respondents to withdraw a patent application, including filings under the Patent Cooperation Treaty, that claimed priority from technology a former laser engineer had access to during his employment with a defence systems group. The applicant secured an interlocutory injunction — urgent relief granted before trial — on the strength of a serious question that the engineer had misused confidential information under section 183 of the Corporations Act.

The strategic lesson sits underneath the drama: the case was won on the paper trail. Expert evidence aligning the internal concept report with the patent claims, a signed contract assigning IP and defining confidential information, and contemporaneous emails reminding the engineer he could not draw on employer IP — that disciplined record is what made the claim credible enough to justify the order.

The takeaway for in-house IP leads, GCs and founders is that who owns employee inventions in Australia is decided long before a dispute, in the IP clauses in employment contracts and the documentation habits around sensitive projects. Three things are worth checking now: that your contracts assign IP and define confidential information in terms that survive public disclosure; that significant R&D leaves a dated, attributable trail linking people to concepts; and that you monitor patent databases for filings by former staff, because timing is decisive — a forced withdrawal can cost a competitor its priority date permanently, which is exactly why the relief had bite here.

The harder truth is that protecting confidential information from employees is a strategy, not a clause. Once an invention is published, the confidentiality is gone for good, so the value of a strong contract and clean records is that they let you act in the narrow window before that happens.

Read the full judgment — EOS Defence Systems Pty Limited v Pahlavani [2026] FCA 688.