Under the shadow of tender investigations, AKSHI becomes the "brain" of artificial intelligence in Albania

The National Agency for the Information Society (AKSHI) is expected to take on the role of the coordinating authority for artificial intelligence in Albania, according to the draft law "On Artificial Intelligence", issued for public consultation by the government.

The draft law aims to create the first complete legal framework for the development, commercialization and use of artificial intelligence systems in the country, fully aligning with European Union Regulation 2024/1689 (AI Act).

In Article 3 of the draft law, the NAASHI is defined as the “coordinating authority/Structure for AI”, while its role in the implementation and monitoring of artificial intelligence systems is also defined. The draft law defines that the “AI Structure” is the function of the authority that “has the task of implementing, monitoring and supervising AI systems and general-purpose AI models”.

According to the draft, the AKSHI will be part of the group of competent authorities together with the Commissioner for the Right to Information and Personal Data Protection, market surveillance authorities and other institutions that may be determined by decision of the Council of Ministers.

Rules also apply to foreign providers, some AI practices are prohibited

The draft law provides for a broad scope of application that extends beyond companies located in Albania. According to Article 2, the law will apply to “providers that place on the market or put into service AI systems or place on the market general-purpose AI models, regardless of whether these providers are established or located within Albania or in another country.”

The draft also provides that the law also applies to providers and users located outside Albania "when the result produced by the AI ​​system is used in Albania".

The draft also establishes a list of prohibited artificial intelligence practices, similar to those provided for in the European Union's AI Act.

These include systems that use manipulative or deceptive techniques to influence the behavior of individuals, systems that exploit people's weaknesses due to age, disability, or economic status, as well as systems of social evaluation of individuals.

It also prohibits the use of systems that create facial recognition databases through the indefinite collection of images from the internet or security cameras.

Another ban relates to the use of artificial intelligence systems to elicit the emotions of individuals in workplaces and educational institutions, except when they are used for medical or security purposes.

Limitations on real-time biometric identification, what about high-risk systems?

The draft law places specific restrictions on the use of real-time biometric identification systems in public spaces.

According to Article 5, their use for law enforcement purposes will be permitted only in special cases, including the search for victims of kidnapping or trafficking, the prevention of serious threats to life and security, and the identification of persons suspected of serious criminal offenses.

The use of these systems will require prior authorization from the competent court. The draft stipulates that “no decision with a negative legal effect on a person may be taken solely based on the result of the real-time biometric identification system.”

The bill creates a special category for high-risk artificial intelligence systems.

These systems will be subject to obligations for risk management, technical documentation, automatic event recording (logs), transparency and post-market monitoring.

The draft requires that for high-risk systems, “a continuous and iterative process” of risk management be established throughout the product life cycle.

Likewise, systems must be developed on databases that are “relevant, sufficiently representative, and as much as possible, error-free and complete,” with measures to identify and mitigate algorithmic biases.

Measures for startups and testing environments

The draft law includes provisions aimed at supporting innovation and the development of new technologies.

It envisages the creation of “regulatory AI sandboxes”, which are defined as controlled environments where providers can develop, train and test innovative artificial intelligence systems under the supervision of authorities.

For small and medium-sized enterprises, including startups, the draft also provides for simplified forms of technical documentation for conformity assessment processes.

In the introduction to the draft law, the government emphasizes that the law is "fully aligned" with Regulation (EU) 2024/1689 of the European Parliament and of the Council, known as the AI ​​Act.

The purpose of the law, according to Article 1, is to promote "the development, commercialization, deployment and use of human-centered and trustworthy artificial intelligence", while guaranteeing a high level of protection for health, safety, fundamental rights, democracy, the rule of law and the environment./ Monitor