szellemi tulajdon (IP)

in English: intellectual property

basicwidespread

Intellectual property is the totality of rights attached to creations of the human mind – works, inventions, brands – which determine who may use and exploit them.

In plain terms

Intellectual property is like ownership of a house, but it applies not to a physical object but to an intellectual creation. It determines who owns a novel, a song, a logo, software, or an invention, and what they can do with it. The fence here is not made of wire but of law.

It serves to make creation and development worthwhile: whoever invests time and money into a work or invention can decide on its use for a time and charge for it. In return, protection is mostly time-limited; afterwards, the work becomes public domain. Trademarks and trade secrets are exceptions: these can theoretically be protected indefinitely if the holder maintains them.

In the age of AI, copyright has made the news because large language and image models were trained on vast amounts of copyrighted text, images and music. Two questions are in dispute: whether it is permissible to use this material for training without permission, and who owns what the machine or an independently working agent produces.

How it works

It consists of several distinct legal areas: copyright protects works (text, image, music, software), patents protect inventions, trademarks protect brand names and logos, industrial designs protect the external appearance of products, and trade secrets protect confidentially managed knowledge.
Copyright arises automatically when a work is created; it does not need to be registered. Patents, trademarks and industrial designs, however, must be applied for at an intellectual property office: in Hungary, the Hungarian Intellectual Property Office; for EU-wide or European protection, the EU Intellectual Property Office (EUIPO) or the European Patent Office.
The rights holder can grant permission (a license) to others for use, free of charge or for a fee, and can take action against anyone who uses their work without permission.
The law also recognizes exceptions, such as quotation, private copying, or text and data mining. In the EU, rights holders can prohibit the use of their works for data mining, including AI training. For works available on the internet, this must be indicated in a machine-readable way. The ban does not apply to data mining for scientific purposes at research institutions.
With AI, this arises at three points: what the model was trained on, whether the output resembles a protected work too closely, and whether what the machine produces without significant human creative work receives any protection at all.

An everyday example

You create a logo for your business with an image generator, and it strikingly resembles a known brand's symbol. If used this way, you may receive a cease-and-desist letter for trademark infringement. If the image is unique but entirely machine-generated, it is quite possible you cannot prevent others from copying it.

Who is affected

entrepreneur, executiveYou need to know who owns AI-generated materials in your company and whether they infringe on others' rights.

marketerBefore using generated images, texts, or music, you must check the provider's license terms and similarity to existing works. → Marketing

developerCode written by a coding agent can also contain snippets covered by open-source licenses, which come with obligations. → Code

parent, student, teacherFor AI content used in essays and teaching materials, it matters where the source comes from and what can be adopted. → Education & Family

A common misconception

MythWhat I create with AI is automatically mine and protected by copyright.

In realityCopyright in most countries requires human creative effort, so content produced purely by machines typically receives no protection. The fact that the provider's contract grants you usage rights to the output is not the same as legal protection.

Limits and risks

Rules differ by country: what falls under fair use in the United States is not necessarily legal in the EU, and vice versa.
In the United States, Germany, and the United Kingdom, the first rulings on AI training have already been issued. However, these partly contradict each other; many cases are still under appeal or settled by agreement, so practice may still change in the coming years.
AI output can unintentionally come too close to a protected work or brand. Liability typically lies with the user. Some service providers assume legal risks arising from this under certain conditions, but the extent varies by provider.
If you upload confidential company material to a public AI service, you risk compromising trade secret protection.

Similar, but not the same

copyright – Copyright is only one branch of intellectual property, protecting literary, artistic, and scientific works, as well as software.
data protection – Data protection safeguards people's personal data, while intellectual property protects creations and the economic interests attached to them.
open source – Open-source software is also protected by copyright. The rights holder attaches a license that allows widespread use, but often ties it to conditions, such as displaying the author's name or sharing the source code.

Related concepts

Broader

AI regulation, legal compliance

Narrower

On the edge

training data, text and data mining, generative AI, EU AI Act, open source

A brief history

The first modern copyright law was the 1710 English Statute of Anne. The foundations of the international system were laid by the 1883 Paris Convention for the Protection of Industrial Property and the 1886 Berne Convention for the Protection of Literary and Artistic Works. The treaty establishing the World Intellectual Property Organization (WIPO) was signed in Stockholm in 1967 and entered into force in 1970.

Where things stand

The concept itself has long been an established legal basis, but its application to AI is still taking shape at the end of 2026: lawsuits over training data are ongoing worldwide. Since August 2025, the EU AI Act has required providers of general-purpose models to apply copyright rules and publish a public summary of the training data.

Status: 2026 November

In the news

Appears in 8 Agent Age stories; the curve shows weekly occurrence in the news magazine.

Cyberattacks on the legal sector double as attackers deploy autonomous AI agents, CrowdStrike saysOct 29, 2026 · Jog
Disney seeks Character.AI assets as states sue the company over minors' self-harmOct 19, 2026 · Education & Family
Harvard and Imperial College create AI avatars of some of their professorsOct 19, 2026 · Education & Family
Law firms turn their own expertise into AI technologyOct 12, 2026 · Jog
Singapore launches public consultation on AI and intellectual property lawSep 28, 2026 · Jog
Insurers split over AI coverage as CFC launches IP cover and Munich Re's aiSure backs live lending modelSep 23, 2026 · Money
Nvidia, Palantir and other firms restrict use of advanced AI models over privacy concernsSep 14, 2026 · Jog
Argentina's Perfil sues OpenAI and Microsoft over use of journalistic contentAug 24, 2026 · Jog

Also known as

szellemi tulajdon · szellemi tulajdonjog · szellemitulajdon-jog · intellectual property rights · IP · intellectual property · IPR · intellectual property rights

Sources

Wikipedia – Intellectual property

This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.