The EU has reached a provisional agreement on its landmark AI Act, originally introduced in 2021, softening some compliance requirements following intensive lobbying by tech giants including Apple, Microsoft, Google, Amazon, and Meta. The revised rules maintain a risk-based framework while reducing burdens on startups. Critically, bans on AI “nudifier” apps, deepfake pornography, and non-consensual intimate imagery are retained, with Article 50 mandating watermarking of AI-generated content. Parliament negotiator Brando Benifei confirmed “strong protections for citizens” remain intact. The agreement now proceeds to final legislative review and vote.
In-Depth:
- The European Union (EU) has finally agreed on a set of artificial innotifyigence (AI) regulations.
- New rules assist developers and startups to comply better.
- To ensure that high-risk artificial innotifyigence systems applications are protected, Law has established safeguards.
- International legislation might influence international standards for AI governance.
EU nereceivediators have receivedten close to completing the formulation of a new law regulating the utilize of artificial innotifyigence, but the current version displays the effects of many months of lobbying by companies operating in the indusattempt for less strict regulation.
Nereceivediations between members representing EU member states and the European Parliament have led to a tentative agreement in Brussels after much discussion, leading to a relaxation of earlier, stricter proposals, but still maintaining stronger regulations for higher-risk systems.
An official from the European Commission stated that “this is a balanced approach that safeguards fundamental rights and drives innovation.” Subsequently, the rules are created public by the European Commission.
scrut.io
How Compliance Rules for EU AI Act Have Changed
EU AI Act, which was introduced in 2021, follows a risk-based approach and sorts AI systems on the basis of how harmful they can be. Previous versions necessitated extensive documentation, transparency reporting, and risk assessment for a wide array of AI systems, including generative AI tools.
The European Union is tightening rules especially around AI-generated nudity, deepfakes and sexually explicit content under its AI Act and related digital laws.
The rules are aimed at tackling revenge porn, deepfake abutilize, child safety risks and misleading AI-generated sexual imagery while forcing tech companies to improve transparency and moderation systems.
These alters were created to assist decrease the burden on start-ups and tinyer firms, without jeopardizing the protection of sensitive sectors. The text of the entire law was not yet officially published, and some aspects may alter before the final adoption of the legal text.
Who lobbied for alter
The legislative effort was subjected to lobbying efforts from large technology companies and indusattempt bodies on both sides of the Atlantic.
Bodies such as Digital Europe, which speaks on behalf of companies like Apple, Microsoft and Google, created submissions on policy positions encouraging the EU legislature to reduce obligations and clarify what constitutes high-risk AI systems. Digital Europe policy positions, Indusattempt group Computer & Communications Indusattempt Association, with members like Amazon and Meta, also lobbied the EU through consultation responses and policy positions for a more nuanced legislation.
The indusattempt lobbied against overly restrictive regulations, which would drive investments and expertise out of Europe, especially into the United States and China. This was recognized by the legislators in Europe, who nevertheless maintained that the final agreement keeps important safeguards intact.
“We have agreed on strong protections for citizens while leaving enough space for technology to develop,” stated Brando Benifei, another member of the nereceivediating team of the Parliament in a statement provided to members of parliament and the press after completing nereceivediations. European Parliament statement .
NGOs expressed worries about potential loosening of safeguards, especially with respect to certain exceptions linked to national security or law enforcement. This was denied by EU officials during subsequent briefings.
Implications for the rest of the world from Europe’s AI legislation
Should it come into effect, there are expected to be significant implications for worldwide technological standards as well. This has been observed before, particularly with the advent of the GDPR in 2018, as many international businesses tconclude to design their regulatory frameworks based on EU laws; a trconclude also known as the “Brussels effect.”
While the AI Act will ban certain applications, such as social scoring and manipulation by AI systems, it will also regulate the utilize of other applications according to the level of risk involved.
Compliance efforts have already begun among businesses as well. A policy executive at a European startup association stated that although things are somewhat better than before, compliance is a difficult matter for startups. “It is good news that the compliance burden is reduced. However, it is still a challenge for startups.”
The provisional agreement now relocates to the final legislative stage, where EU institutions will review and vote on the completed text.
![[INTERVIEW] Korea, EU expand research alliance amid tech race, geopolitical shifts](https://foundernews.eu/storage/2026/05/1778287647_2fc3daf8-46c5-463c-964b-dfc078117078.jpg)
