The EU AI Act is the world’s first comprehensive legal framework for artificial intelligence (AI). It is a set of rules that apply across the European Union to make sure artificial intelligence is used safely. Its goal is to support the development of AI that is trustworthy and focused on people, while also protecting things like health, safety, basic rights, democracy, the rule of law, and the environment.
Given companies’ increased desire to use AI to drive efficiencies, particularly through Generative AI (GenAI), businesses must lay the foundations to implement AI in a responsible and controlled manner.
The EU AI Act makes it clear that AI must be properly managed throughout its use. Organisations need to know where AI is being used, what risks it may create, and who is responsible for overseeing it.
What is the AI Act?
The EU AI Act groups AI systems into different risk levels. Most AI systems are considered low risk, so they don’t have to meet any special requirements under the Act. However, the higher the risk, the stricter the rules that apply.
This means organisations need to understand what AI systems they use and decide which risk category each one falls into. The risk level will determine what rules the organisation needs to follow, while also protecting your organisation, reduces legal risk.
It is a risk-based approach, and defines 4 levels of risk for AI systems:
- Unacceptable
- High-risk
- Limited risk
- Minimal risk
Unacceptable Risk:
All AI systems considered a clear threat to the safety, livelihoods and rights of people are banned.
Examples include: (1) harmful AI-based manipulation and deception and (2) untargeted scraping of the internet or CCTV material to create or expand facial recognition databases.
The European Commission has published prohibited AI artificial intelligence (AI) practices, as defined by the AI Act to help businesses comply with these regulations. You can find the guidelines here.
High Risk:
AI use cases that can pose serious risks to health, safety or fundamental rights are classified as high-risk. These include: (1) AI tools for employment, management of workers and access to self-employment (e.g. CV-sorting software for recruitment) and AI solutions used in education institutions, that may determine the access to education and course of someone’s professional life (e.g. scoring of exams).
High risk AI systems will need to follow strict regulations and they can be found here.
Limited Risk:
This refers to the risks associated with a need for transparency around the use of AI. The AI Act introduces specific disclosure obligations to ensure that humans are informed when necessary to preserve trust. For instance, when using AI systems such as chatbots, humans should be made aware that they are interacting with a machine so they can take an informed decision.
Moreover, providers of generative AI have to ensure that AI-generated content is identifiable. On top of that, certain AI-generated content should be clearly and visibly labelled, namely deep fakes and text published with the purpose to inform the public on matters of public interest.
Minimal risk:
The vast majority of everyday AI tools fall into this group and standard data privacy (GDRP) and cybersecurity rules apply. They are often automated, low-stakes tasks, and do not involve profiling or decision-making. Examples can include: (1) Software that sorts junk mail from your inbox and (2) Grammar and spelling assistants or predictive text.
How do I comply with the AI Act in my organisation?
According to the Government of Ireland, organisation should follow these steps to comply with the AI Act across all organisations.
- Identify where AI is used in your organisation
- Determine the risk category of the AI system
- Create internal policies for responsible AI use
- Build AI literacy within the organisation by upskilling your employees with free training through FIT’s AI Training Centre of Excellence
Next Steps for your Organisation:
Failing to follow the rules can have serious consequences. Organisations can face large fines of up to €35 million or 7% of their total worldwide annual turnover of the preceding year, whichever is higher, as well as investigations by regulators. There is also the risk of reputational damage and real-world harm if AI systems make mistakes or are used irresponsibly.
For organisations who wish to upskill their employees in AI, FIT (Fastrack into Information Technology) have partnered with global leaders Google, Microsoft, IBM, Intel and Accenture to provide free, self-paced AI training. The training spans from introductory to advanced levels and can support organisations in building internal AI capacity.
Learn more about FIT’s AI Training Centre of Excellence here or contact the team on info@fit.ie today.
