August 25, 2025 • 22 mins
Artificial intelligence is rapidly becoming central to areas such as public health, education, agriculture, and climate resilience. In this context, the role of the State is coming into sharper focus, particularly in how governments can shape innovation to serve broad social goals. Intellectual property frameworks, often seen as tools for exclusivity, are being repurposed to support inclusive access and public benefit.
This special episode of Intangiblia was recorded as part of my participation in the workshop “The Role of the State in Advancing Equitable Access to AI,” taking place in Oxford in September 2025. Organized by Sumaya Nur Adan and Joanna Wiaterek, and supported by the Future of Life Institute, the event brings together legal scholars, policymakers, and technologists to examine how States can ensure that the benefits of AI are equitably shared.
The episode explores five legal and policy mechanisms that are already influencing how AI is governed through intellectual property. It discusses Canada’s ongoing efforts to map and license Crown-owned patents under a broader national strategy. It examines Singapore’s copyright reforms, which have introduced clear legal exceptions to support AI model training. The conversation also includes examples of culturally aware AI development, such as the open-source Falcon model in the UAE and community-led Indigenous data initiatives in New Zealand. It looks at how public interest licensing and voluntary IP pools are evolving in fields beyond health, and how state-led initiatives, such as public procurement and open research mandates, are being used to align technological development with social needs.
The episode also reviews recent legal rulings in the United States that have tested the limits of fair use in AI training. These include the 2024 decision involving OpenAI, the 2025 dismissal of claims against Meta, and the Bartz v. Anthropic case presided over by Judge Alsup, which underscored the difference between statistical pattern recognition and direct reproduction of copyrighted works.
Rather than focusing solely on restrictions or incentives, the discussion emphasizes how IP law can serve as a strategic governance tool. By adapting legal frameworks to current challenges, States can guide AI innovation toward inclusive outcomes and help ensure that technological advancement remains aligned with the public good.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Artemisa (00:00):
Artificial intelligence isn't just about
chips and code.
It's about who gets to shapethe future and who gets left out
.
From health breakthroughs toclimate tools, AI can serve the
public good, but only if thesystems around it, like
intellectual property, aredesigned for access, not just
ownership.
So what happens when IP becomesa bridge instead of a barrier?
(00:20):
Stick around?
We're about to unpack howstates around the world are
making that leap.
Barrier Stick around.
Speaker 2 (00:25):
We're about to unpack how states around the world are
making that leap.
You are listening toIntangiblia, the podcast of
intangible law, playing talkabout intellectual property.
Please welcome your host,Leticia Caminero.
Leticia Caminero (AI) (00:45):
Welcome to Intangiblia, where we explore
the invisible threads thatconnect ideas, innovation and
law.
I'm Leticia Caminero, and thisis a special bonus episode to
mark my participation in theworkshop the Role of the State
in Advancing Equitable Access toAI happening this September in
Oxford.
Organized by Sumayya Nour Adanand Joanna Weaterek, this
(01:07):
inaugural workshop is dedicatedto exploring how governments can
operationalize the benefits ofartificial intelligence,
ensuring equitable access forall.
The event is supported by theFuture of Life Institute, a
global organization committed tosteering transformative
technology toward the publicgood.
In this episode, we'll explorehow states are rethinking
(01:30):
intellectual property tools tomake AI more open, ethical and
inclusive, from open sourcelicensing to copyright reforms,
public innovation, procurementand more.
Before we get into thetechnical parts, let's talk
about the real reason we're here.
It's not just about whatgovernments are doing with AI.
It's about why they're doing itExactly.
Artemisa (01:50):
Everyone's talking about access to AI tools, but
access isn't enough if thosetools aren't built for or by the
people who need them.
Leticia Caminero (AI) (01:58):
That's why the role of the state is so
crucial, not as a passiveregulator, but as an active
enabler, funding, negotiatingand designing the systems that
shape innovation outcomes.
Artemisa (02:12):
And let's be honest, if systems weren't originally
built with equity in mind, theywere designed to protect
exclusive rights and attractprivate investment.
Leticia Caminero (AI) (02:22):
But AI is changing the game.
It's making innovation faster,more data-driven and way more
dependent on shared resourceslike datasets, infrastructure
and collective knowledge.
Artemisa (02:35):
Which means the IP toolkit needs to evolve, not to
throw out incentives, but tostretch the system so that
innovation also serves people,not just profits.
Leticia Caminero (AI) (02:45):
In this episode, we're going to unpack
five smart strategiesgovernments are using to do
exactly that.
These aren't just abstractideas.
They're real policy tools withlegal teeth.
Artemisa (02:58):
We're talking public interest, licensing, copyright
reforms, open source, ai andprocurement strategies that
promote digital public goods.
So if you've ever wondered howlaw can shape fairness in the
age of algorithms, stay tuned.
Leticia Caminero (AI) (03:13):
Before we dive in, a quick note about how
this episode was made.
Artemisa (03:18):
As your AI co-host.
I was generated usingartificial intelligence tools my
voice, my personality, evensome of my sass crafted with
code, and Leticia's voice today,also cloned with AI.
Leticia Caminero (AI) (03:33):
The script you're hearing is a blend
of human research and machinesupport.
We used AI to organize legalframeworks, summarize case law
and shape the narrative, butevery word was fact-checked and
refined by a real human.
That's me Well real me.
Artemisa (03:49):
So, yes, this is an episode about AI, made with AI,
but always with transparency,accountability and a healthy
respect for the law.
Leticia Caminero (AI) (03:58):
Now let's explore how governments are
rewriting the rules ofinnovation, starting with how
they license the technologiesthey fund.
Let's start with something thatsounds obvious but often isn't
when the public pays forinnovation, the public should
(04:20):
benefit from it.
Artemisa (04:22):
Right.
Too often taxpayer-fundedresearch ends up locked behind
patents, paywalls or proprietarycode that nobody can touch
without a license or a lawyer.
Leticia Caminero (AI) (04:32):
But that's beginning to change.
Some governments are nowattaching conditions to public
R&D funding to make sure theintellectual property generated
is shared or at least licensed.
More broadly.
Artemisa (04:46):
Take Canada's Explore IP strategy.
It maps out government-fundedIP and encourages licensing
across sectors.
Some provinces are evenconsidering models where
crown-owned patents must be madeavailable for public interest
uses, especially in sectors likehealth, climate and AI interest
(05:06):
uses, especially in sectorslike health, climate and AI.
Leticia Caminero (AI) (05:07):
Meanwhile countries like the Netherlands,
germany and New Zealand arepiloting open supply and open
sunlight mandates for AI outputscreated with public funding.
These include data sets, modelsand training tools designed to
stay open by default.
Artemisa (05:23):
And this isn't just policy wonk stuff.
It's how you build digitalpublic goods AI tools that solve
real problems, from floodprediction to rural healthcare,
that any country or startup canreuse or adapt.
The upside faster innovation,especially in developing
countries or underserved regions.
Lower costs for researchers andsmall businesses.
(05:43):
Transparency in modeldevelopment and small businesses
.
Transparency in modeldevelopment and training data.
Public trust in how AI isfunded and deployed.
Leticia Caminero (AI) (05:51):
The challenge is, of course, there's
tension.
Some inventors or universitiesworry that open licensing may
reduce the commercial value oftheir discoveries.
Artemisa (06:02):
Or that mandatory openness could discourage
private investment or slow downtech transfer.
Leticia Caminero (AI) (06:07):
That's why many experts now recommend a
flexible hybrid model keepingcore IP accessible for public
use while still allowingexclusive licenses in specific
contexts.
Artemisa (06:22):
Bottom line.
When the state funds AI, it hasthe power and the
responsibility to make sure theresults don't gather dust in a
patent vault.
Leticia Caminero (AI) (06:32):
Or worse, get bought up by a private firm
and turn into an access barrier.
Let's make that publicinvestment count.
When it comes to AI, copyrightis one of the most contested
frontiers.
After all, ai models learn fromexisting content books, images,
articles, music but is that?
Artemisa (06:53):
legal.
Enter the world of copyrightexceptions for text and data
mining, or TDM.
These exceptions allow machinesto analyze large amounts of
copyrighted content withoutasking for permission.
Each time, at least in someplaces.
Leticia Caminero (AI) (07:09):
The European Union's DSM directive
introduced a structured approach.
It allows non-commercial TDM bydefault and permits commercial
TDM if rights holders don't optout.
Artemisa (07:25):
Meanwhile, singapore has gone further.
In 2021, it passed aprogressive copyright reform
that explicitly allows datamining for both commercial and
non-commercial AI training, withno opt-out clause.
Leticia Caminero (AI) (07:38):
Japan and the UK have also carved out TDM
exceptions.
The US, however, relies on amore flexible but less
predictable concept.
Artemisa (07:48):
fair use In 2024, a US federal judge dismissed part of
a high-profile lawsuit againstOpenAI ruling that using
publicly available content totrain AI can qualify as fair use
.
Then came June 2025, a busymonth for AI and copyright.
First, meta won a similarlawsuit brought by a group of
(08:09):
authors, including PulitzerPrize winner Michael Chabon.
A US judge sided with thecompany, stating that the
authors failed to prove thatMeta's Lama models reproduced
their copyrighted books in anymeaningful way.
Leticia Caminero (AI) (08:22):
The ruling emphasized that using
large volumes of text to extractstatistical patterns without
direct copying or replacing themarket could fall under fair use
.
It's strengthening the notionthat not all ingestion equals
infringement.
Artemisa (08:40):
with another opinion in Barts v Anthropic PBC, he
dismissed most of the claims,stating that model training does
not necessarily violatecopyright if the outputs do not
substantially resemble thesource works.
Leticia Caminero (AI) (08:57):
Judge Alassip went further, warning
that overextending copyright lawto cover mere learning by
machines could chill innovation.
He stressed that copyrightprotects expression, not facts
or functional analysis, and thatanthropic outputs had to be
judged on what they actuallyproduced, not just what they
(09:17):
were trained on.
Artemisa (09:19):
It's early days and higher courts might still weigh
in, but taken together, the openAI, meta and anthropic rulings
give AI developers in the US acautiously optimistic roadmap,
especially when working withpublic content and ensuring
non-verbatim outputs.
Leticia Caminero (AI) (09:37):
The White House AI Action Plan, published
in early 2025, acknowledges thetension.
It calls for clearer guidelinesand multi-stakeholder dialogue
on IP rights in AI training,particularly around data sets,
transparency and ethicalboundaries.
Artemisa (09:58):
And while the federal government hasn't proposed
sweeping reforms yet, it'spromoting sector-specific
initiatives like encouragingopen licensing for publicly
funded data sets and supportingthe National AI Research
Resource, which includes shareddata and computing tools for
researchers.
Leticia Caminero (AI) (10:16):
So in the US the direction is clear.
The government sees value inmaking AI development more
inclusive and transparent, andthat includes rethinking how
copyright law supports or blocksthat mission.
But there's a catch Even iftraining qualifies as fair use,
many models are so opaque thatwe can't tell what copyrighted
(10:40):
material they ingested or howit's being used.
Artemisa (10:44):
This is known as the black box problem.
Without transparency intraining data and model behavior
, it's hard to assess compliance, bias or accountability.
Leticia Caminero (AI) (10:56):
That's why some jurisdictions are now
linking TIRMAC exceptions totransparency obligations.
Think documentation, data set,registries or even watermarking
outputs.
Artemisa (11:09):
It's a balancing act between enabling innovation and
protecting creators, betweenbuilding powerful models and
understanding how they think.
Leticia Caminero (AI) (11:17):
Copyright law is evolving.
The question is whether itevolved fast enough and fairly
enough to guide AI towardssocially beneficial outcomes.
Artemisa (11:29):
The tools are there, the policies are emerging.
Now it's about building systemsthat serve both human and
machine learning.
Speaker 2 (11:39):
Intangiblia, the podcast of intangible law.
Playing talk about intellectualproperty.
Leticia Caminero (AI) (11:46):
Let's talk about open source AI, not
just as a technical choice, butas a strategy for inclusion.
Artemisa (11:54):
In places where private AI development is
limited or foreign techdominates the market, open
source models can be a lifeline.
They let communities build,adapt and own the tools they
need.
Leticia Caminero (AI) (12:05):
They let communities build, adapt and own
the tools they need.
We've seen this clearly in theUAE, where the Technology
Innovation Institute releasedFalcon, an open source large
language model.
Artemisa (12:24):
The goal help researchers and developers in
Arabic speaking countries workwith high performing models
tailored to their region, and inNew Zealand, the
government-backed Te Heku Mediaproject is using open-source AI
to protect and revive Maorilanguage and culture.
They're training models onIndigenous data sets with full
community consent to createspeech recognition and text
tools grounded in local values.
Leticia Caminero (AI) (12:43):
This is a powerful reminder.
Inclusive AI isn't just aboutaccess.
It's about relevance AI thatreflects local languages,
knowledge systems and socialrealities.
Artemisa (12:56):
Open source licensing helps make that possible.
It removes commercial and legalbarriers, accelerates
localization and invitesgrassroots innovation.
Leticia Caminero (AI) (13:05):
And when the state gets involved by
funding, curating or deployingthese models, it multiplies the
impact.
We move from isolatedinnovation to scalable public
infrastructure.
Artemisa (13:20):
Challenges to watch Data quality and bias.
Even open models need strongdata sets and ethical guidelines
.
Capacity gaps.
Local teams may need support,not just code.
Sustainability maintaining openmodels requires long-term
funding and stewardship.
What works?
Successful programs ofteninclude community participation,
(13:42):
transparent licensing terms andclear government leadership.
Leticia Caminero (AI) (13:48):
And the IP structures behind them, like
Creative Commons, open dataagreements or open source model
cards, are what keep the doorsopen for future use.
Artemisa (13:59):
Bottom line.
When AI tools are built foreveryone, they work better for
everyone.
Leticia Caminero (AI) (14:05):
And open source is one way states are
putting that idea into practice.
Now let's talk about somethingthat sounds very technical but
is actually very powerful IPpooling and public interest
licensing.
Artemisa (14:20):
These are tools states can use to negotiate access
instead of just regulating orreacting, and, when done right,
they allow governments to shareintellectual property across
sectors, companies or evencountries.
Leticia Caminero (AI) (14:33):
A perfect example is the COVID-19
Technology Access Pool, or CTAP,created by the World Health
Organization.
It invited patent holders tovoluntarily license health
technologies for broader,low-cost access.
Artemisa (14:51):
While CTAP didn't attract as many tech
contributors as hoped, the ideabehind it is starting to gain
traction in AI.
Leticia Caminero (AI) (15:01):
We're seeing more discussion around
sovereign patent pools for AImodels and data sets, especially
those created with publicfunding.
This means governments canconsolidate certain IPS sets and
license them non-exclusivelyfor high-impact applications
like education, agriculture orpublic health.
Artemisa (15:24):
This approach is flexible.
It keeps the door open forprivate sector engagement, but
with terms that reflect equity,like requiring licensees to
serve underserved markets ordisclose how they use.
The model Otis unlockscross-border collaboration on AI
for development.
Lower licensing barriers forsmall businesses and NGOs.
Lower licensing barriers forsmall businesses and NGOs.
(15:45):
Stronger negotiation power forstates in global AI deals.
Leticia Caminero (AI) (15:51):
Of course , voluntary licensing only works
when there's trust andtransparency, and IP pools
require solid infrastructure,clear governance and incentive
structures that work for bothrights holders and the public.
Artemisa (16:04):
But here's the real innovation treating IP as a
negotiable asset, not just alegal right, something that can
be structured to support access,adaptation and scaling, not
just exclusivity.
Leticia Caminero (AI) (16:19):
In a world where AI systems often
cross borders and sectors,public interest licensing gives
governments the tools to stay inthe game and shape outcomes
that benefit more people.
Artemisa (16:32):
Especially when private innovation doesn't
automatically serve the publicgood.
Leticia Caminero (AI) (16:38):
Let's wrap up the strategy walkthrough
with a big idea Governments notjust reacting to innovation,
but leading it.
Artemisa (16:46):
We mean governments acting like innovators
themselves, using publicprocurement, challenge funds and
sandbox environments to steerAI in directions that serve
society.
Leticia Caminero (AI) (16:56):
Exactly, this is the space of
government-led open innovation,and it's getting traction fast
From national AI sandboxes todata collaboratives and AI
research hubs.
States are creating controlspaces where risks are managed
and equities designed in.
Artemisa (17:15):
Let's take the G7s AI Grand Challenges.
These are publicly fundedmissions designed to crowd in
innovation for the public good,focusing on trustworthy AI,
healthcare and sustainability.
They don't just ask for privatesolutions.
Leticia Caminero (AI) (17:29):
They co-design the terms of access
and IP and the EuropeanCommission's AI factories
combine funding, infrastructureand open licensing frameworks to
accelerate collaborative modeldevelopment for small firms and
researchers.
Artemisa (17:43):
What makes it open?
Pre-competitive collaboration,ip terms baked into contracts,
not as an afterthought.
Shared data infrastructures,public oversight of outputs.
Leticia Caminero (AI) (17:52):
What's tricky.
States need legal capacity tobuild this.
That means smart contractdesign, ip literacy and
long-term digital governance,not just flashy announcements.
Artemisa (18:07):
And they must avoid extractive models where public
R&D gets privatized at thefinish line.
The goal is to keep innovationcirculating, not captured.
Leticia Caminero (AI) (18:16):
This is the future States as co-creators
, not just regulators, notblocking innovation, but
unlocking it on terms thatinclude more people and more
possibilities.
Artemisa (18:26):
And using IP law as a lever, not a wall.
Leticia Caminero (AI) (18:30):
So what can policymakers actually do
with all this?
We've explored five big ideas,and now it's time to draw some
conclusions.
Artemisa (18:40):
Not slogans, not theory practical steps.
If you're a government actor,innovation funder or policymaker
, here's a roadmap to make AImore inclusive using the power
of IP code datasets includeclear licensing terms that
promote broad reuse.
Leticia Caminero (AI) (18:55):
Not everything has to be open source
, but public money should createpublic value Two codify fair
training practices.
Artemisa (19:15):
introduce or clarify copyright exceptions for text
and data mining withtransparency requirements,
consider sector-specificcarve-outs or guidelines for
ethical AI training, especiallyin education, health and public
language models.
Three support regional andcultural relevance.
Fund open source models andtools built for and by local
(19:36):
communities.
Prioritize languages,indigenous data and
domain-specific AI that reflectthe realities of diverse
populations and protectcontributors through clear
benefit sharing and datasovereignty frameworks.
Leticia Caminero (AI) (19:51):
Four enable voluntary IP sharing
mechanisms.
Build and maintain publicinterest, patent pools or open
licensing platforms.
Offer financial or reputationalincentives for rights holders
who contribute.
Ensure clear governance andalignment with national
development goals.
Artemisa (20:12):
Five lead innovation by example.
Use public procurement andgovernment-led challenges to
drive inclusive outcomes.
Design IP terms from the start.
Ensuring that AI built withpublic involvement remains
accessible, safe and auditable.
Leticia Caminero (AI) (20:27):
All of these tools already exist.
The real shift is in how wecombine them with courage,
creativity and a long-term view.
Artemisa (20:38):
Because equitable access to AI isn't just about
what's invented.
It's about what's shared,what's protected and who gets to
use it.
Leticia Caminero (AI) (20:46):
And that means rewriting the rules, not
to limit progress but to invitemore people in.
That's a wrap on this specialbonus episode of Indangibria.
Whether you're a policymaker,researcher, legal advisor or
simply curious about the futureof AI and society researcher,
legal advisor or simply curiousabout the future of AI and
(21:07):
society we hope this gave younew ways to think about how IP
can support, not stifle,equitable innovation.
Artemisa (21:11):
From open licensing to copyright exceptions, from
community-driven models topublic innovation strategies,
we've seen that the role of thestate isn't just about catching
up to AI.
It's about shaping it.
Leticia Caminero (AI) (21:23):
This episode was created to accompany
my participation the realLeticia that is in the workshop
the Role of the State inAdvancing Equitable Access to AI
, organized by Sumaya Nooradanand Joanne Weaterek with support
from the Future of LifeInstitute.
We'll be back soon with ourregular season of interviews,
(21:45):
inventions and imagination.
Artemisa (21:47):
Until then, keep questioning, keep creating and,
above all, keep innovating withintention.
Speaker 2 (21:58):
Intangiblia the podcast of intangible law plain
talk about intellectual property.
Did you like what we talkedtoday?
Please share with your network.
Do you want to learn more aboutintellectual property?
Subscribe now on your favoritepodcast player.
Follow us on Instagram,Facebook, linkedin and Twitter.
Visit our website,wwwintangibliacom.
(22:22):
Copyright Leticia Caminero 2020.
All rights reserved.
This podcast is provided forinformation purposes only.
Artificial intelligence isn't just about
chips and code.
It's about who gets to shapethe future and who gets left out
.
From health breakthroughs toclimate tools, AI can serve the
public good, but only if thesystems around it, like
intellectual property, aredesigned for access, not just
ownership.
So what happens when IP becomesa bridge instead of a barrier?
(00:20):
Stick around?
We're about to unpack howstates around the world are
making that leap.
Barrier Stick around.
Speaker 2 (00:25):
We're about to unpack how states around the world are
making that leap.
You are listening toIntangiblia, the podcast of
intangible law, playing talkabout intellectual property.
Please welcome your host,Leticia Caminero.
Leticia Caminero (AI) (00:45):
Welcome to Intangiblia, where we explore
the invisible threads thatconnect ideas, innovation and
law.
I'm Leticia Caminero, and thisis a special bonus episode to
mark my participation in theworkshop the Role of the State
in Advancing Equitable Access toAI happening this September in
Oxford.
Organized by Sumayya Nour Adanand Joanna Weaterek, this
(01:07):
inaugural workshop is dedicatedto exploring how governments can
operationalize the benefits ofartificial intelligence,
ensuring equitable access forall.
The event is supported by theFuture of Life Institute, a
global organization committed tosteering transformative
technology toward the publicgood.
In this episode, we'll explorehow states are rethinking
(01:30):
intellectual property tools tomake AI more open, ethical and
inclusive, from open sourcelicensing to copyright reforms,
public innovation, procurementand more.
Before we get into thetechnical parts, let's talk
about the real reason we're here.
It's not just about whatgovernments are doing with AI.
It's about why they're doing itExactly.
Artemisa (01:50):
Everyone's talking about access to AI tools, but
access isn't enough if thosetools aren't built for or by the
people who need them.
Leticia Caminero (AI) (01:58):
That's why the role of the state is so
crucial, not as a passiveregulator, but as an active
enabler, funding, negotiatingand designing the systems that
shape innovation outcomes.
Artemisa (02:12):
And let's be honest, if systems weren't originally
built with equity in mind, theywere designed to protect
exclusive rights and attractprivate investment.
Leticia Caminero (AI) (02:22):
But AI is changing the game.
It's making innovation faster,more data-driven and way more
dependent on shared resourceslike datasets, infrastructure
and collective knowledge.
Artemisa (02:35):
Which means the IP toolkit needs to evolve, not to
throw out incentives, but tostretch the system so that
innovation also serves people,not just profits.
Leticia Caminero (AI) (02:45):
In this episode, we're going to unpack
five smart strategiesgovernments are using to do
exactly that.
These aren't just abstractideas.
They're real policy tools withlegal teeth.
Artemisa (02:58):
We're talking public interest, licensing, copyright
reforms, open source, ai andprocurement strategies that
promote digital public goods.
So if you've ever wondered howlaw can shape fairness in the
age of algorithms, stay tuned.
Leticia Caminero (AI) (03:13):
Before we dive in, a quick note about how
this episode was made.
Artemisa (03:18):
As your AI co-host.
I was generated usingartificial intelligence tools my
voice, my personality, evensome of my sass crafted with
code, and Leticia's voice today,also cloned with AI.
Leticia Caminero (AI) (03:33):
The script you're hearing is a blend
of human research and machinesupport.
We used AI to organize legalframeworks, summarize case law
and shape the narrative, butevery word was fact-checked and
refined by a real human.
That's me Well real me.
Artemisa (03:49):
So, yes, this is an episode about AI, made with AI,
but always with transparency,accountability and a healthy
respect for the law.
Leticia Caminero (AI) (03:58):
Now let's explore how governments are
rewriting the rules ofinnovation, starting with how
they license the technologiesthey fund.
Let's start with something thatsounds obvious but often isn't
when the public pays forinnovation, the public should
(04:20):
benefit from it.
Artemisa (04:22):
Right.
Too often taxpayer-fundedresearch ends up locked behind
patents, paywalls or proprietarycode that nobody can touch
without a license or a lawyer.
Leticia Caminero (AI) (04:32):
But that's beginning to change.
Some governments are nowattaching conditions to public
R&D funding to make sure theintellectual property generated
is shared or at least licensed.
More broadly.
Artemisa (04:46):
Take Canada's Explore IP strategy.
It maps out government-fundedIP and encourages licensing
across sectors.
Some provinces are evenconsidering models where
crown-owned patents must be madeavailable for public interest
uses, especially in sectors likehealth, climate and AI interest
(05:06):
uses, especially in sectorslike health, climate and AI.
Leticia Caminero (AI) (05:07):
Meanwhile countries like the Netherlands,
germany and New Zealand arepiloting open supply and open
sunlight mandates for AI outputscreated with public funding.
These include data sets, modelsand training tools designed to
stay open by default.
Artemisa (05:23):
And this isn't just policy wonk stuff.
It's how you build digitalpublic goods AI tools that solve
real problems, from floodprediction to rural healthcare,
that any country or startup canreuse or adapt.
The upside faster innovation,especially in developing
countries or underserved regions.
Lower costs for researchers andsmall businesses.
(05:43):
Transparency in modeldevelopment and small businesses
.
Transparency in modeldevelopment and training data.
Public trust in how AI isfunded and deployed.
Leticia Caminero (AI) (05:51):
The challenge is, of course, there's
tension.
Some inventors or universitiesworry that open licensing may
reduce the commercial value oftheir discoveries.
Artemisa (06:02):
Or that mandatory openness could discourage
private investment or slow downtech transfer.
Leticia Caminero (AI) (06:07):
That's why many experts now recommend a
flexible hybrid model keepingcore IP accessible for public
use while still allowingexclusive licenses in specific
contexts.
Artemisa (06:22):
Bottom line.
When the state funds AI, it hasthe power and the
responsibility to make sure theresults don't gather dust in a
patent vault.
Leticia Caminero (AI) (06:32):
Or worse, get bought up by a private firm
and turn into an access barrier.
Let's make that publicinvestment count.
When it comes to AI, copyrightis one of the most contested
frontiers.
After all, ai models learn fromexisting content books, images,
articles, music but is that?
Artemisa (06:53):
legal.
Enter the world of copyrightexceptions for text and data
mining, or TDM.
These exceptions allow machinesto analyze large amounts of
copyrighted content withoutasking for permission.
Each time, at least in someplaces.
Leticia Caminero (AI) (07:09):
The European Union's DSM directive
introduced a structured approach.
It allows non-commercial TDM bydefault and permits commercial
TDM if rights holders don't optout.
Artemisa (07:25):
Meanwhile, singapore has gone further.
In 2021, it passed aprogressive copyright reform
that explicitly allows datamining for both commercial and
non-commercial AI training, withno opt-out clause.
Leticia Caminero (AI) (07:38):
Japan and the UK have also carved out TDM
exceptions.
The US, however, relies on amore flexible but less
predictable concept.
Artemisa (07:48):
fair use In 2024, a US federal judge dismissed part of
a high-profile lawsuit againstOpenAI ruling that using
publicly available content totrain AI can qualify as fair use
.
Then came June 2025, a busymonth for AI and copyright.
First, meta won a similarlawsuit brought by a group of
(08:09):
authors, including PulitzerPrize winner Michael Chabon.
A US judge sided with thecompany, stating that the
authors failed to prove thatMeta's Lama models reproduced
their copyrighted books in anymeaningful way.
Leticia Caminero (AI) (08:22):
The ruling emphasized that using
large volumes of text to extractstatistical patterns without
direct copying or replacing themarket could fall under fair use
.
It's strengthening the notionthat not all ingestion equals
infringement.
Artemisa (08:40):
with another opinion in Barts v Anthropic PBC, he
dismissed most of the claims,stating that model training does
not necessarily violatecopyright if the outputs do not
substantially resemble thesource works.
Leticia Caminero (AI) (08:57):
Judge Alassip went further, warning
that overextending copyright lawto cover mere learning by
machines could chill innovation.
He stressed that copyrightprotects expression, not facts
or functional analysis, and thatanthropic outputs had to be
judged on what they actuallyproduced, not just what they
(09:17):
were trained on.
Artemisa (09:19):
It's early days and higher courts might still weigh
in, but taken together, the openAI, meta and anthropic rulings
give AI developers in the US acautiously optimistic roadmap,
especially when working withpublic content and ensuring
non-verbatim outputs.
Leticia Caminero (AI) (09:37):
The White House AI Action Plan, published
in early 2025, acknowledges thetension.
It calls for clearer guidelinesand multi-stakeholder dialogue
on IP rights in AI training,particularly around data sets,
transparency and ethicalboundaries.
Artemisa (09:58):
And while the federal government hasn't proposed
sweeping reforms yet, it'spromoting sector-specific
initiatives like encouragingopen licensing for publicly
funded data sets and supportingthe National AI Research
Resource, which includes shareddata and computing tools for
researchers.
Leticia Caminero (AI) (10:16):
So in the US the direction is clear.
The government sees value inmaking AI development more
inclusive and transparent, andthat includes rethinking how
copyright law supports or blocksthat mission.
But there's a catch Even iftraining qualifies as fair use,
many models are so opaque thatwe can't tell what copyrighted
(10:40):
material they ingested or howit's being used.
Artemisa (10:44):
This is known as the black box problem.
Without transparency intraining data and model behavior
, it's hard to assess compliance, bias or accountability.
Leticia Caminero (AI) (10:56):
That's why some jurisdictions are now
linking TIRMAC exceptions totransparency obligations.
Think documentation, data set,registries or even watermarking
outputs.
Artemisa (11:09):
It's a balancing act between enabling innovation and
protecting creators, betweenbuilding powerful models and
understanding how they think.
Leticia Caminero (AI) (11:17):
Copyright law is evolving.
The question is whether itevolved fast enough and fairly
enough to guide AI towardssocially beneficial outcomes.
Artemisa (11:29):
The tools are there, the policies are emerging.
Now it's about building systemsthat serve both human and
machine learning.
Speaker 2 (11:39):
Intangiblia, the podcast of intangible law.
Playing talk about intellectualproperty.
Leticia Caminero (AI) (11:46):
Let's talk about open source AI, not
just as a technical choice, butas a strategy for inclusion.
Artemisa (11:54):
In places where private AI development is
limited or foreign techdominates the market, open
source models can be a lifeline.
They let communities build,adapt and own the tools they
need.
Leticia Caminero (AI) (12:05):
They let communities build, adapt and own
the tools they need.
We've seen this clearly in theUAE, where the Technology
Innovation Institute releasedFalcon, an open source large
language model.
Artemisa (12:24):
The goal help researchers and developers in
Arabic speaking countries workwith high performing models
tailored to their region, and inNew Zealand, the
government-backed Te Heku Mediaproject is using open-source AI
to protect and revive Maorilanguage and culture.
They're training models onIndigenous data sets with full
community consent to createspeech recognition and text
tools grounded in local values.
Leticia Caminero (AI) (12:43):
This is a powerful reminder.
Inclusive AI isn't just aboutaccess.
It's about relevance AI thatreflects local languages,
knowledge systems and socialrealities.
Artemisa (12:56):
Open source licensing helps make that possible.
It removes commercial and legalbarriers, accelerates
localization and invitesgrassroots innovation.
Leticia Caminero (AI) (13:05):
And when the state gets involved by
funding, curating or deployingthese models, it multiplies the
impact.
We move from isolatedinnovation to scalable public
infrastructure.
Artemisa (13:20):
Challenges to watch Data quality and bias.
Even open models need strongdata sets and ethical guidelines
.
Capacity gaps.
Local teams may need support,not just code.
Sustainability maintaining openmodels requires long-term
funding and stewardship.
What works?
Successful programs ofteninclude community participation,
(13:42):
transparent licensing terms andclear government leadership.
Leticia Caminero (AI) (13:48):
And the IP structures behind them, like
Creative Commons, open dataagreements or open source model
cards, are what keep the doorsopen for future use.
Artemisa (13:59):
Bottom line.
When AI tools are built foreveryone, they work better for
everyone.
Leticia Caminero (AI) (14:05):
And open source is one way states are
putting that idea into practice.
Now let's talk about somethingthat sounds very technical but
is actually very powerful IPpooling and public interest
licensing.
Artemisa (14:20):
These are tools states can use to negotiate access
instead of just regulating orreacting, and, when done right,
they allow governments to shareintellectual property across
sectors, companies or evencountries.
Leticia Caminero (AI) (14:33):
A perfect example is the COVID-19
Technology Access Pool, or CTAP,created by the World Health
Organization.
It invited patent holders tovoluntarily license health
technologies for broader,low-cost access.
Artemisa (14:51):
While CTAP didn't attract as many tech
contributors as hoped, the ideabehind it is starting to gain
traction in AI.
Leticia Caminero (AI) (15:01):
We're seeing more discussion around
sovereign patent pools for AImodels and data sets, especially
those created with publicfunding.
This means governments canconsolidate certain IPS sets and
license them non-exclusivelyfor high-impact applications
like education, agriculture orpublic health.
Artemisa (15:24):
This approach is flexible.
It keeps the door open forprivate sector engagement, but
with terms that reflect equity,like requiring licensees to
serve underserved markets ordisclose how they use.
The model Otis unlockscross-border collaboration on AI
for development.
Lower licensing barriers forsmall businesses and NGOs.
Lower licensing barriers forsmall businesses and NGOs.
(15:45):
Stronger negotiation power forstates in global AI deals.
Leticia Caminero (AI) (15:51):
Of course , voluntary licensing only works
when there's trust andtransparency, and IP pools
require solid infrastructure,clear governance and incentive
structures that work for bothrights holders and the public.
Artemisa (16:04):
But here's the real innovation treating IP as a
negotiable asset, not just alegal right, something that can
be structured to support access,adaptation and scaling, not
just exclusivity.
Leticia Caminero (AI) (16:19):
In a world where AI systems often
cross borders and sectors,public interest licensing gives
governments the tools to stay inthe game and shape outcomes
that benefit more people.
Artemisa (16:32):
Especially when private innovation doesn't
automatically serve the publicgood.
Leticia Caminero (AI) (16:38):
Let's wrap up the strategy walkthrough
with a big idea Governments notjust reacting to innovation,
but leading it.
Artemisa (16:46):
We mean governments acting like innovators
themselves, using publicprocurement, challenge funds and
sandbox environments to steerAI in directions that serve
society.
Leticia Caminero (AI) (16:56):
Exactly, this is the space of
government-led open innovation,and it's getting traction fast
From national AI sandboxes todata collaboratives and AI
research hubs.
States are creating controlspaces where risks are managed
and equities designed in.
Artemisa (17:15):
Let's take the G7s AI Grand Challenges.
These are publicly fundedmissions designed to crowd in
innovation for the public good,focusing on trustworthy AI,
healthcare and sustainability.
They don't just ask for privatesolutions.
Leticia Caminero (AI) (17:29):
They co-design the terms of access
and IP and the EuropeanCommission's AI factories
combine funding, infrastructureand open licensing frameworks to
accelerate collaborative modeldevelopment for small firms and
researchers.
Artemisa (17:43):
What makes it open?
Pre-competitive collaboration,ip terms baked into contracts,
not as an afterthought.
Shared data infrastructures,public oversight of outputs.
Leticia Caminero (AI) (17:52):
What's tricky.
States need legal capacity tobuild this.
That means smart contractdesign, ip literacy and
long-term digital governance,not just flashy announcements.
Artemisa (18:07):
And they must avoid extractive models where public
R&D gets privatized at thefinish line.
The goal is to keep innovationcirculating, not captured.
Leticia Caminero (AI) (18:16):
This is the future States as co-creators
, not just regulators, notblocking innovation, but
unlocking it on terms thatinclude more people and more
possibilities.
Artemisa (18:26):
And using IP law as a lever, not a wall.
Leticia Caminero (AI) (18:30):
So what can policymakers actually do
with all this?
We've explored five big ideas,and now it's time to draw some
conclusions.
Artemisa (18:40):
Not slogans, not theory practical steps.
If you're a government actor,innovation funder or policymaker
, here's a roadmap to make AImore inclusive using the power
of IP code datasets includeclear licensing terms that
promote broad reuse.
Leticia Caminero (AI) (18:55):
Not everything has to be open source
, but public money should createpublic value Two codify fair
training practices.
Artemisa (19:15):
introduce or clarify copyright exceptions for text
and data mining withtransparency requirements,
consider sector-specificcarve-outs or guidelines for
ethical AI training, especiallyin education, health and public
language models.
Three support regional andcultural relevance.
Fund open source models andtools built for and by local
(19:36):
communities.
Prioritize languages,indigenous data and
domain-specific AI that reflectthe realities of diverse
populations and protectcontributors through clear
benefit sharing and datasovereignty frameworks.
Leticia Caminero (AI) (19:51):
Four enable voluntary IP sharing
mechanisms.
Build and maintain publicinterest, patent pools or open
licensing platforms.
Offer financial or reputationalincentives for rights holders
who contribute.
Ensure clear governance andalignment with national
development goals.
Artemisa (20:12):
Five lead innovation by example.
Use public procurement andgovernment-led challenges to
drive inclusive outcomes.
Design IP terms from the start.
Ensuring that AI built withpublic involvement remains
accessible, safe and auditable.
Leticia Caminero (AI) (20:27):
All of these tools already exist.
The real shift is in how wecombine them with courage,
creativity and a long-term view.
Artemisa (20:38):
Because equitable access to AI isn't just about
what's invented.
It's about what's shared,what's protected and who gets to
use it.
Leticia Caminero (AI) (20:46):
And that means rewriting the rules, not
to limit progress but to invitemore people in.
That's a wrap on this specialbonus episode of Indangibria.
Whether you're a policymaker,researcher, legal advisor or
simply curious about the futureof AI and society researcher,
legal advisor or simply curiousabout the future of AI and
(21:07):
society we hope this gave younew ways to think about how IP
can support, not stifle,equitable innovation.
Artemisa (21:11):
From open licensing to copyright exceptions, from
community-driven models topublic innovation strategies,
we've seen that the role of thestate isn't just about catching
up to AI.
It's about shaping it.
Leticia Caminero (AI) (21:23):
This episode was created to accompany
my participation the realLeticia that is in the workshop
the Role of the State inAdvancing Equitable Access to AI
, organized by Sumaya Nooradanand Joanne Weaterek with support
from the Future of LifeInstitute.
We'll be back soon with ourregular season of interviews,
(21:45):
inventions and imagination.
Artemisa (21:47):
Until then, keep questioning, keep creating and,
above all, keep innovating withintention.
Speaker 2 (21:58):
Intangiblia the podcast of intangible law plain
talk about intellectual property.
Did you like what we talkedtoday?
Please share with your network.
Do you want to learn more aboutintellectual property?
Subscribe now on your favoritepodcast player.
Follow us on Instagram,Facebook, linkedin and Twitter.
Visit our website,wwwintangibliacom.
(22:22):
Copyright Leticia Caminero 2020.
All rights reserved.
This podcast is provided forinformation purposes only.
Popular Podcasts
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Clay Travis and Buck Sexton Show
The Clay Travis and Buck Sexton Show. Clay Travis and Buck Sexton tackle the biggest stories in news, politics and current events with intelligence and humor. From the border crisis, to the madness of cancel culture and far-left missteps, Clay and Buck guide listeners through the latest headlines and hot topics with fun and entertaining conversations and opinions.