The European Union of Amnesty International aims to create a flat stadium to innovate artificial intelligence: here is what it is

European Union AI Law, KnownHe was Described By the European Commission as “the first comprehensive law in the world”. After years of making, it has gradually became part of the reality of 450 million people living in the 27 countries that include the European Union.

However, the European Union of Amnesty International is more than just a European relationship. It applies to both local and foreign companies, and can affect both service providers and the spread of artificial intelligence systems; The European Commission cents examples of how it is applied to the CV developer and to the bank that buys this tool. Now all these parties have a legal framework that paves the way for its use of artificial intelligence.

Why is the European Union Act of Amnesty International?

As usual, with the European Union’s legislation, the European Union Law is Amnesty International to ensure that there is a unified legal framework that applies to a specific topic across the European Union countries – this time is Amnesty International. Now that the list is in place, it should “ensure freedom of movement, across the border, for the goods and services based on artificial intelligence” without divergence of local restrictions.

With a timely manner, the European Union seeks to create a level stadium throughout the region Civilization confidenceWhich can also create opportunities for startups. However, the common framework that I adopted is not fully lenient: despite the relatively early stage of artificial intelligence adopting widely in most sectors, the European Union’s International Union Law puts a big tape of what artificial intelligence should do and should not do it for society on a broader scale.

What is the purpose of the European Union’s law for Amnesty International?

According to European legislators, the main goal of the framework is to “enhance the absorption of central and reliable artificial intelligence while ensuring a high level of health and safety protection and basic rights as stipulated in the European Union’s basic rights charter, including democracy, the state of law, environmental protection, protection of harmful effects of AI in the Union, and support for innovation.

Yes, this is very mouth, but it is worth the analysis carefully. First, because many will depend on how to define “center” and “worthy of confidence” from artificial intelligence. Second, because it gives a good feeling of risky balance to maintain different goals: innovation in exchange for preventing damage, as well as absorbing AI in exchange for environmental protection. As usual, with the legislation of the European Union, again, Satan will be in detail.

How does the European Union act to Amnesty International to balance its various goals?

To achieve a balance between the damage against the potential benefits of Amnesty International, the European Union law has adopted artificial intelligence a Risk -basedBring a handful of “unacceptable risk” cases; A set of “high risk” uses the call for tight regulation; And apply lighter obligations to “limited risk” scenarios.

Has the European Union law entered into force?

Yes and no. The European Union AI Law began on August 1, 2024, but it will not come into effect except through a series of Overlooking dates for intertwined. In most cases, it will also be applied to new arrivals from companies that already provide artificial intelligence products and services in the European Union.

He entered the deadline on the deadline on February 2, 2025, and focused on enforcing the ban on a small number of prohibited uses of Amnesty International, such as the uniform spy of the Internet or CCTV for facial images to raise or expand databases. Many others will follow, but unless the schedule changes, most provisions will be applied by mid -2016.

What changed on August 2, 2025?

Since August 2, 2025, the European Union AI Law has applied to “artificial intelligence models for systemic purposes with systemic risks.”

GPAI models (AI for general purposes) are Amnesty International Models trained with a large amount of data, and can be used for a wide range of tasks. This is where the risk element comes. According to the European Union Law, GPAI models can come Regular risk – “For example, by reducing barriers to the development of chemical or biological weapons, or unintended issues to control the models of autonomy (GPAI).”

Before the deadline, the European Union was published Guidelines For service providers from GPAI models, which include both European companies and non -European players such as anthropor, Google, Meta, Openai. But since these companies already have models on the market, they will also get them until August 2, 2027, unlike the new arrivals.

Is the European Union AI law its teeth?

The European Union Law comes from artificial intelligence with sanctions that legislators wanted to be “effective, proportional and vital” – even for great international players.

Details will be placed by the European Union countries, but the regulation determines the total spirit – that the sanctions will differ according to the level of risk that are considered – as well as thresholds for each level. Violation over prohibited artificial intelligence requests leads to the highest penalty of up to “up to 35 million euros or 7 % of the total annual rotation rate all over the world for the previous fiscal year (whichever is higher).”

European Commission can also enroll fines of up to 15 million euros or 3 % of annual sales rotation on GPAi models.

How long does current players intend to comply?

the GPAI volunteer practice blogIncluding obligations such as not training models on pirated content, is a good indication of how to deal with companies with the framework law in order to be forced to do so.

In July 2025, dead She announced that he will not sign the GPAI volunteer practice blog What is meant to help these service providers comply with the European Union AI law. However, Google shortly after He confirmed that he will signDespite the reservations.

Signers To date, it includes Aleph Alpha, Amazon, Anthropic, COHERE, Google, IBM, Microsoft, Mistral AI, Openai and others. But as we saw with the example of Google, the signature is not completely equal.

Why (some) technology companies fight these rules?

While saying in a Blog post Google will sign GPAI volunteer practices blog, its head for global affairs, Kent Walker, still has reservations. “We are still concerned that the law of artificial intelligence and the risks of code slows the development of Europe and spread artificial intelligence.”

Meta was more extremist, where the chief international affairs official Joel Kaplan said in A. mail On LinkedIn, “Europe is heading to the wrong way to artificial intelligence.” He stated that the practice code “is called the European Union’s implementation of the AI“ transgression ”that the practice code“ provides a number of legal uncertainty for moderators, as well as measures that exceed the scope of the AI law. ”

European companies have also expressed their concerns. Arthur Minche, CEO of the French AI, Mistral Ai, was part of a group of European CEOs who An open message signed In July 2025, Brussels urged Ali “Stop around the clock” For two years, before the main obligations of the European Union law entered into artificial intelligence.

Will the schedule change?

In early July 2025, the European Union responded negatively to pressure the efforts calling for a temporary stop, saying it will continue to adhere to its schedule to implement the European Union law. Go on August 2, 2025, the deadline as planned, and we will update this story if anything changes.