Future of Creative Work

Creative Protections in the Age of AI

Introduction: Creative Data and Ownership

From paintbrushes to Adobe Illustrator, tools have been the cornerstone of creative work. So, what makes AI creative tools any different? Why is the use of Artificial Intelligence controversial within the creative sector? One word: data.

The Center for Security and Emerging Technologies explains the important role of data to the development of generative Artificial Intelligence in just 13 words: “Machine learning systems use computing power to execute algorithms that learn from data.” This data that machine learning systems use to train models are often public datasets scrapped from the internet. These public datasets can include copyright creative works that has since lapsed (A.A. Milne’s Winnie the Pooh has no copyright protections since 2022) and may also include still copyrighted protected works (For example: the protected Disney version of the red shirt wearing Winnie the Pooh wherein Disney earns money through licensing).

The Issues

Now this is where creatives, AI developers, and policymakers are on a standstill: AI developers see this publicly scraped data as ‘fair use‘ pointing out that they are not passing creative works as their own but rather just getting statistical data from it. Creatives such as the Authors’ Guild in the US see the works scrapped by generative AI to be competing with them as creatives on a commercial standpoint… in real time. As the CEO of the Authors’ Guild put it: AI companies are being valued for billions of dollars, meanwhile the artist community does not. Meanwhile, policymakers in the UK have gone through controversial consultations in amending the Copyright law that seeks a middle ground between creatives and developers. Global discourse weighs heavy in balancing the importance of being ahead of technological innovations (the AI race) while protecting the human-in-the-loop aspect of creative work. 

The Need for Public Awareness

Do we wait for these protective laws and discourse to be resolved before engaging with AI tools? No, that sets back innovation and creative exploration of human creativity. Nevertheless, we should be aware of how these tools can change the way we compensate and publish creative work. Even if AI companies haven’t reached ‘human level intelligence yet’, these AI tools have already irreversible long-lasting impacts especially in the creative sector of which copyright laws are now being reviewed.  

As we wait for these developments on data, ownership and policy, knowing the basics of creative rights in the age of AI for non-legal practitioners is essential while the rapid development of this technology continues and creatives continue to self-publish on the World Wide Web.

 

This article aims to introduce to creatives and common people the intersection of technology with creative ownership in a non-legalese friendly approach, and to inform policymakers and creative advocates with cases that can influence creative industries policies overtime. This article is divided in 3 parts, 1) Creative Technology and Protections in the lens of history; 2) Definitions of Intellectual Property; and 3) Governing Methods for Creative Work of Frontier AI Technology Companies. This article is accompanied by an interactive map highlighting developments that can influence AI governance and Intellectual Property, and a practical guide for creatives in using AI tools.

This page will be updated accordingly in an attempt to keep up with the developments. 

The Industrial Revolutions and Birth of Creative Protections

The Industrial Revolution brought lasting changes and economic transformation via new technologies such as the steam engine and electricity. Due to technology, education, and access to capital, the many Industrial Revolutions prompted the rise of income of individual persons across the world through the creation of new work and industries.

As work and industries evolve, governance and interventions have also been created in response to protect human work aided by technology. The cottage industries (such as weaving) did not need as much governance compared to when it evolved to factory textile work (aided by the spinning Jenny)– trade unions came into rise during this time promoting better working conditions.

Manual block printing evolved into the mechanized printing press. Over time, these manual text blocks developed into a new form of creative work with the innovation of computing during the Third Industrial Revolution. Through the legal intervention of licensing, type design—the creation of fonts and typefaces—has become an industry involving producers, foundries (designers), and distributors (such as Adobe and Monotype) around the world. New technological innovations also created the industries of photography and film. Below is a set of innovations per Industrial Revolution and the industrial governance that comes with it.

Quite simply: New technological innovations create new tools for creative work. Creative work generates income through creative ownership. Creative ownership, in turn, leads to protections and regulations that safeguard and promote human creative value.

With the emergence of AI as a new technological tool, how are creative work, creative rights and governance being redefined? A starting point of this discussion is one of the primary ways that creative ownership has been protected throughout history— the system of Intellectual Property.

Creative Rights Challenged: Intellectual Property in the Age of AI

Intellectual Property is the cornerstone of creatives and creative industries. Intellectual Property refers to original creative creations of the mind. IP is protected by law “which enable people to earn recognition or financial benefit from what they invent or create”.The World Intellectual Property Office has created an explainer ‘ How to Make a Living in the Creative Industries’ page that summarizes the ways intellectual property is embedded in the various creative industries all over the world.  

Examples of intellectual property as defined by WIPO are: 

Trademark: “A trademark is a sign capable of distinguishing the goods or services of one enterprise from those of other enterprises. Trademarks date back to ancient times when artisans used to put their signature or “mark” on their products.”

Industrial Design: “An industrial design constitutes the ornamental or aesthetic aspect of an article. A design may consist of three-dimensional features, such as the shape or surface of an article, or of two-dimensional features, such as patterns, lines or color.”

Patents:”A patent is an exclusive right granted for an invention. Generally speaking, a patent provides the patent owner with the right to decide how – or whether – the invention can be used by others. In exchange for this right, the patent owner makes technical information about the invention publicly available in the published patent document.”

Geographical Indicators: “Geographical indications and appellations of origin are signs used on goods that have a specific geographical origin and possess qualities, a reputation or characteristics that are essentially attributable to that place of origin.”

Copyright: “Copyright is a legal term used to describe the rights that creators have over their literary and artistic works. Works covered by copyright range from books, music, paintings, sculpture and films, to computer programs, databases, advertisements, maps and technical drawings.”

Among these types of intellectual property, copyright is the one most directly challenged by the development of artificial intelligence. In most countries, copyright protection is automatically granted upon the creation of an original work, without the need for a formal registration process (although stronger protections are available through official registration).

In this context, data scraping— the automated process of extracting large amounts of information from websites, databases, or online platforms using software, bots, or scripts — raises serious concerns. Scraped data may include copyrighted materials collected without the permission of their original creators.

These scraped datasets serve as the “fuel” for training large language models (LLMs) — AI models that require extensive data to perform tasks accurately. For example, to produce a reliable image of a dog, a model typically needs to be trained on approximately 150–500 images per category to achieve good accuracy.

The quality, provenance, and ethics of these training datasets — especially in relation to frontier AI companies like OpenAI and DeepSeek — are among the main concerns raised by creatives today, as AI models continue to evolve at a rapid pace.

How Frontier AI Companies are responding to Creative Protection

In response to these concerns (and several intellectual property legal battles), frontier AI companies are adopting a range of approaches to engage with the creative community. While not all of these approaches seem sufficient across different industries, these agile responses and adjustments to their models are important to monitor, especially as creative intellectual property rights continue to be discussed and debated.

With the constant evolution and introduction of various creative AI tools, the table below summarizes the creative concerns related to the three AI companies at the forefront of disrupting the creative industries. The tables below aims to provide creatives and policymakers with insight into the methods of acquiring data sets, inputs and outputs of these creative AI tools, as well as each company’s response to creative ownership. Some categories are based on The Allan Turing Institute’s Center for Emerging Technology and Security paper on Strengthening Resilience to AI Risk. Selected categories are found relevant in assessing the development of select creative AI tools, this table will be updated accordingly.

DESIGN, TRAINING AND TESTING

	DeepSeek (Janus)	OpenAI (Dall-E, Sora)	RunwayML	StabilityAI (Stable Diffusion)	Meta (Llama)
Direct Engagement with Industry Bodies / Public Figures Engagements of frontier AI companies with creatives and creative communities		Public figures who wish for their depiction not to be generated can inform OpenAI through a how-to article⁠(opens in a new window).		Spawning AI worked with StabilityAI in giving artists the opportunity to opt out training data sets.
Training Data Methods Methods use in acquiring data for model training		DallE have an open call out for Open Source Training Data Sets and Private Data Sets.	Inputs and Outputs may be used by the Company to train and improve its AI models.
Model Reporting: Training Data Sets Information on data sets used for model training	.	DallE was trained with sets of text and image pairs, source unknown. OpenAI and Microsoft funded Harvard University’s Free AI Training Data set of public domain books.		Trained on the LAION-5B and Common Crawl (used by Creative Commons) datasets. Further discussion soon.	Alleged (case ongoing) to have used scraped data from Library Genesis, a digital warehouse of stolen intellectual property.

DEPLOYMENT AND USAGE

	DeepSeek (Janus)	OpenAI (Dall-E, Sora)	RunwayML	StabilityAI (Stable Diffusion)	Meta (Llama)
Commercial Rights – Output Information on who own the commercial rights of generated outputs	DeepSeek grants users ownership rights in outputs generated by the services.	Users own the commercial rights of generated output on Dall-E.	Users own the commercial rights of generated output on Runway.	Selected paid tiers are given commercial rights of generated outputs.
Opt Out – Model Training Availability of opting out inputs and outputs in training AI models	Input and Output:   DeepSeek reserves the right to use inputs and outputs to “provide, maintain, operate, develop or improve” its services . There are no available opt outs for model training.	Output only:   Users can opt out the content being generated to train OpenAI models. A guide can be seen here.	There are no available information on opt outs for model training.	Input only: Users were given a chance to opt out, this ended in March 2023.	Input only: As of 2024, EU users can opt out of Meta’s model training which will be using everyone’s social media posts.

 

This table will updated overtime. Updated as of April 27, 2025.

What now? A responsibility in crafting futures.

With all of these information, what is there to do? 

Though lawyers all over the world are already watching every legal case that can take as a precedent to copyright claims, the work does not stop (or start) with legal professionals. Creatives themselves should be responsible to keep up (at the very least be aware) with the intersection of AI developments, policies and governance surrounding the creative industries and new tools. 

Meanwhile, policymakers and creative organizations with the capacity to engage in discussions on changes to protections and policies for creatives should also understand the concerns surrounding the development of new technologies as creative tools, without sacrificing the rights of those practicing in the field.

Academics and innovators have also joined in creating more fair futures between humans and machines. University of Chicago created The Glaze Project, a research efforts that develops technical tools for creatives in the age of generative AI. A MIT Technology Review’s Innovator of the Year, the Glaze Project consists of tools such as Glaze, that can be used on every piece of artwork artists post online to protect themselves, and Nightshade an optional feature that can be used to deter unscrupulous model trainers.

These kind of innovations and participation in the field of AI is essential by keeping abreast of developments in policies and governance surrounding the creative industries and technology. Effective governance is formed through a multi-stakeholder approach of the private and public sector in establishing a conducive environment wherein humans and machine are encouraged to be innovative and creative. From private creative individuals to AI developers and the academe, cross-sector collaboration is key to crafting equitable creative futures.

Future of Creative Work – Map
For policymakers and creative organisations

This interactive map is a companion piece of this Future of Creative work article to provide a global perspective on evolving challenges and opportunities of AI developments . This map highlights developments and cases from around the world that can influence both AI governance and the intellectual property rights of creatives. This page will be updated accordingly. Access the Future of Creative Work map here.

Practical Guide for Creatives on using Creative AI tools 

For creatives

Before using any AI tool that involves your original work, how do you know if the creative work is yours? For further discussion, a Practical Guide to using Creative AI Tools will be available by May 2025.