AI Glossary

A structured set of controls from which an organization selects those that match the risks of a particular AI system and the requirements of several frameworks at once.

A compliance method in which an organization implements a single set of controls that meets the requirements of multiple frameworks at once, thereby avoiding duplication of work.

Naming the author or source of content used or generated by AI; connected with the protection of copyright and with transparency about the origin of content.

A part of the code for all providers of GPAI models that sets out measures to comply with EU copyright law, including respecting rights reservations (opt-out) for text and data mining.

A term for practices that obscure the origin of protected content used for training, so that the output appears original and free of third-party rights.

An expression of will by a rightholder reserving that their work may not be used for text and data mining; it must be expressed in a machine-readable manner.

The automated collection of protected content from the internet for the purpose of training models, which may infringe copyright and website terms of use.

Copyrighted works used in the development or training of an AI model.

A collection of an organization's internal documents and knowledge that an AI system draws on when answering, so that its outputs are based on verified company information.

A step taken to eliminate an identified non-compliance or risk associated with an AI system.

A modernized convention on the protection of individuals with regard to the automated processing of personal data, following on from the original Convention No. 108 of 1981. It responds to profiling and algorithmic processing and introduces, among other things, the right not to be subject to a decision based solely on automated processing.

The first international treaty on artificial intelligence focused on protecting human rights, democracy and the rule of law. It requires ratification by a set number of states in order to enter into force.

A dataset composed of works available under Creative Commons licenses, used for training models under the conditions of the particular license.

An open framework for building systems with multiple cooperating AI agents that divide roles and tasks between them when tackling more complex assignments.

A development environment with built-in AI that helps write, edit and explain code directly while programming.

An event that compromises the availability, integrity, confidentiality or authenticity of data or systems.

The ability of an organization to prevent and recover from cyberattacks.

An EU regulation laying down cybersecurity requirements for products with digital elements throughout their lifecycle; it also applies to software and AI components built into such products.

A body involved in the system of oversight and conformity assessment in the field of AI in the Czech Republic.

The body to which the Czech draft legislation is to entrust the operation of the national regulatory sandbox for artificial intelligence.

OpenAI's text-to-image generation model, one of the pioneering tools of generative visual AI.

AI-supported user interface design techniques that deceptively push users toward decisions against their own interests.

An EU regulation governing access to and sharing of data generated by connected devices, which also affects the availability of data for training and operating AI.

We summarize what you should not overlook in our article Data Act – what you should not overlook.

The voluntary sharing of data by individuals or organizations in the public interest, for example for research, as regulated by the Data Governance Act.

The infrastructure essential for training and running models; its energy demands and location are becoming the subject of political and regulatory debate.

An entity that determines the purposes and means of the processing of personal data.

A change in the characteristics of input data over time that can reduce the reliability of a model without the model itself changing.

A set of rules and processes for managing data within an organization, ensuring its quality, availability, security and legal compliance throughout its lifecycle.

An EU regulation promoting trustworthy data sharing, laying down rules for data intermediaries and the re-use of public sector data.

An entity that controls data generated by a connected product or service and decides on making it available under the rules of the Data Act.