Key Terms and Definitions
- Artificial Intelligence (AI): In the field of computer science, AI is defined as machines and technologies performing tasks previously attributed to humans, such as language processing, decision-making, and identifying objects and analysing data. 1
- AI accountability and advocacy: For the purpose of this research, this refers to civil society efforts to advance protection of human rights in the deployment and regulation of AI as well as to hold deployers and developers accountable for rights violations.
- AI hallucinations: When a large language model (LLM) generates a response “that seems plausible but is actually inaccurate or misleading”. 2
- Algorithmic opacity: The lack of transparency about the factors influencing the decisions and outputs of AI systems.
- AI sandbox: An isolated testing environment where developers and researchers can experiment with AI and trial new models.
- Data annotation: The process of labelling raw data such as images, text, videos, and audio to make it understandable to a machine learning model. 3
- Deepfake: A type of hyper-realistic synthetic media created using GenAI.
- Deep Packet Inspection (DPI): A method of examining and managing network traffic, whereby network packets (smaller segments of information sent over the internet such as an email, video or image) are evaluated. Based on what the content of a packet is, traffic can be altered, blocked or filtered. 4
- Facial Recognition Technology (FRT): A type of AI-powered biometric identification system that identifies or verifies someone’s identity by analysing their facial features.
- Fine-tuning: Databricks defines fine-tuning as “the process of adapting or supplementing pretrained models by training them on smaller, task-specific datasets”. 5
- Generative AI (GenAI): A type of AI that uses algorithms to create and generate synthetic content such as text, videos, photos and audio based on training data patterns. 6
- Human-in-the-oop (HITL): A process or a system where humans are actively involved in AI workflows so that their feedback can help improve models “to ensure accuracy, safety, accountability or ethical decision-making” of AI systems. 7
- Large Language Models (LLMs): In natural language processing, LLMs are AI models trained on large text datasets with the capacity to “understand and generate human-like text, recognize patterns in language, and perform a wide variety of language tasks without task-specific training. They represent a significant advancement in the field of natural language processing (NLP)”. 8
- Machine Learning (ML): A subfield of AI focused on training algorithms on large datasets to learn from data and make their own rules without being programmed. 9
- Misinformation and disinformation: While misinformation is generally defined as the spread of inaccurate information, disinformation is the spread of false information with the intention to deceive or cause harm. 10
- Natural Language Processing: (NLP): A subfield of computer science and linguistics concerned with understanding and analysing human language. Widely deployed in many of the technologies and software people use on a daily basis, NLP combines statistical models, machine learning, and linguistic rules to perform a wide range of tasks including sentiment analysis, machine translation, and text classification. 11
- Phishing: A tactic of crafting and spreading deceiving messages to trick recipients into clicking on a malicious link, downloading an infected file or inputting sensitive personal information such as passwords and financial information. 12
- Spyware: A type of malicious software (malware) that infects devices to collect personal information and monitor user activities, usually on behalf of governments. 13
- Training datasets: Collections of information in different formats—text, images, videos, audio— on which a machine learning model is trained “to make predictions, recognize patterns or generate content”. 14
1 Notre Dame Learning. “AI Overview and Definitions”. Undated. https://learning.nd.edu/resource-library/ai-overview-and-definitions/ (accessed 30 April 2026).
2 Choi, A. & M, K.X. “What are AI hallucinations? Why AIs sometimes make things up”. The Conversation.21 March 2025. https://theconversation.com/what-are-ai-hallucinations-why-ais-sometimes-make-things-up-242896 (accessed 30 April 2026).
3 ​​Coursera. “What Is Data Annotation in Machine Learning”? 19 December 2025. https://www.coursera.org/articles/data-annotation-in-machine-learning (30 April 2026).
4 Abrougui, A. (2025). Phishing, spyware, and smart city tech: Surveillance in Sisi’s Egypt. In T. Roberts & A. Mare (Ed.). Digital Surveillance in Africa: Power, Agency, and Rights (pp. 57–84). London,: Zed Books. Retrieved April 30, 2026, from http://dx.doi.org/10.5040/9781350422117.ch-3
5 Databricks. “What is Fine Tuning”? Undated. https://www.databricks.com/blog/what-is-fine-tuning (30 April 2026).
6 Notre Dame Learning. “AI Overview and Definitions”. Undated. https://learning.nd.edu/resource-library/ai-overview-and-definitions/ (accessed 30 April 2026).
7 IBM. “What is human-in-the-loop”? Undated. https://www.ibm.com/think/topics/human-in-the-loop (30 April 2026).
8 Hugging Face. “Natural Language Processing and Large Language Models”. Undated. https://huggingface.co/learn/llm-course/en/chapter1/2 (accessed 30 April 2026).
9 AI Myths.”The term AI has a clear meaning”. Undated. https://www.aimyths.org/the-term-ai-has-a-clear-meaning (accessed 30 April 2026)
10 United Nations. “Countering Disinformation”. Undated. https://www.un.org/en/countering-disinformation (accessed 30 April 2026).
11 ISO. “Unravelling the secrets of natural language processing”. https://www.iso.org/artificial-intelligence/natural-language-processing (accessed 30 April 2026).
12 Abrougui, A. “Global Trends in Digital Security: Civil Society & Media”. October 2023. Internews. https://internews.org/wp-content/uploads/2023/11/Global-Trends-in-Digital-Security.pdf (accessed 30 April
13 Abrougui, A. “Global Trends in Digital Security: Civil Society & Media”. October 2023. Internews. https://internews.org/wp-content/uploads/2023/11/Global-Trends-in-Digital-Security.pdf (accessed 30 April 2026).
14 IBM. “What is training data”. Undated. https://www.ibm.com/think/topics/training-data (accessed 30 April 2026).