AI governance saw significant activity globally in June 2025, with major developments across the EU, the UK, and the US. Here’s a roundup of the most noteworthy global regulatory news.
European Union:
Delay in implementation: It has been reported that there is a possible delay in the implementation of the EU Artificial Intelligence Act (EU AI Act). The tech lobbying group, CCIA Europe, whose members include Alphabet, Meta, and Apple, on Thursday 26 June 2025, urged the EU to pause the implementation of the EU AI Act, saying that a “rushed roll-out risks jeopardising the continent’s AI aspirations.” The Swedish Prime Minister has called the AI rules “confusing”, and asked the EU to pause its implementation, especially since the guidelines and standards are still pending.
In this vein, the key deadline of 02 August 2025, on which date the rules in the EU AI Act relating to General-Purpose AI (GPAI) Models come into effect, is drawing closer. The GPAI Code of Practice was originally due in May, and the final version is expected to be endorsed and published this week (i.e. the first week of July). A workshop on the Code of Practice is being held on 02 July 2025 for GPAI providers, and the closing plenary, during which the final version of the Code of Practice will be presented, is scheduled to be held on 03 July 2025.
Consultation on classification of high-risk AI systems: A consultation of the classification of AI systems as high-risk is open for six weeks from the 06 June 2025 until the 18 July 2025. The link to the consultation is here.
Germany – DeepSeek: The German Commissioner for Data Protection and Freedom of Information reported DeepSeek and asked Google and Apple to remove the app from their app stores in Germany stating that “the transfer of user data by DeepSeek to China is unlawful.” The Commissioner went on to state that: “DeepSeek has not been able to provide my office with convincing evidence that data of German users is protected in China at a level equivalent to that of the European Union. Chinese authorities have extensive access rights to personal data held by Chinese companies. In addition, DeepSeek users in China do not have enforceable rights and effective legal remedies as guaranteed in the European Union. I have therefore informed Google and Apple, as operators of the largest app platforms, of the violations and expect a prompt review of a blocking.”
European AI Gigafactories initiative: On a different note, it is apt the mention the “overwhelming” response received, with 76 respondents expressing interest in the initiative, far exceeding the Commission’s expectations.
United Kingdom:
AI legislation: In June 2025, the UK government announced the delay in the introduction of AI legislation by at least one year. The UK seems to be re-calibrating its AI regulatory strategy, with the goal of balancing innovation and safety of AI. The bill is expected to include comprehensive provisions on AI’s use of copyrighted material, as tensions on copyright issues escalated over AI training on creative works. Another reason put forward for the delay were concerns that it might weaken the UK’s attractiveness to AI companies, a view which is aligned with that of Donald Trump’s administration.
USA:
Copyright cases: Linked to the increasing concerns on copyright, two contrasting orders were delivered in the US in June.
Meta won an AI copyright lawsuit, in which authors claimed that Meta used books, without the authors’ permission, in order to train its AI model. Although the Judge ruled that the use of copyrighted work without permission would be unlawful “in many circumstances”, the Judge concluded that the authors did not present enough evidence to prove that Meta’s AI model would dilute the market for the authors’ work to show that Meta’s conduct was illegal. In this order however, it is important to note that the Judge highlights that fair use requires “case-by-case analysis” since it is a “fact-specific doctrine.”
This is in contrast with a separate ruling, in which a different Judge concluded that Anthropic’s AI training made fair use of copyrighted materials.
The Judge compared Anthropic model’s use of books to a “reader aspiring to be a writer” who uses previous works as inspiration, not to replicate, but to “create something different.”
The Judge however denied summary judgment on Anthropic’s download of over 7 million pirated books, which were stored in a central library. This was deemed to have infringed the authors’ copyrights and did not therefore constitute fair use under copyright law. The Judge stated that the fact that Anthropic later “bought a copy of a book it stole off the internet will not absolve it of liability for the theft, but it may affect the extent of statutory damages.” A trial to decide on the issue of the pirated copies used to create Anthropic’s central library will be held.
June 2025 also featured a push toward federal AI legislation, as states continued advancing their own rules.
Key developments included:
- Federal moratorium of state AI laws: Congress advanced a provision to bar states from enforcing most AI regulations for the next 10 years. On 30 June 2025, two Republican senators agreed to revise the federal moratorium to five years, as well as permit individual states to implement laws pertaining to the safety of children online and the protection of artists’ images / likeness;
- New York’s RAISE Act: the New York State Legislature passed the Responsible AI Safety and Education (RAISE) Act on 12 June 2025. This bill, if signed into law, would require developers of large-scale AI platforms to implement rigorous safety and security protocols, annual risk reviews, and safeguards to prevent critical harm from AI misuse.
- California: on 02 June 2025, the California Assembly passed the AI Transparency Act, which would require large online platforms to label AI-generated content and require devices to offer digital signatures for authentic images / audio. Meanwhile, the AI Abuse Protection Act passed the California Senate and moved to Assembly committees as of 09 June 2025. This would extend fraud and impersonation laws to cover deepfake images / videos of individuals.
Other International Developments:
G7 Summit – “AI for Prosperity” Declaration: Leaders of the G7 nations issued a joint “G7 Leaders’ Statement on AI for Prosperity” that laid out a collective vision for harnessing AI’s benefits while managing its risks. As part of this, the G7 launched an AI Adoption Roadmap to accelerate the use of AI in the public sector and small-and-medium enterprises (SMEs) across member countries.
Need help staying ahead of the curve with AI regulation? Contact Blue Arrow today, and we can assist you with our sector-specific expertise.
