Here’s what’s on our plate today:

• 🏛️ EU AI Act on hold—how the extra runway could super-charge Big Tech’s lead.

• 📊 Delay it, tweak it, or hit “publish” — have your say.

• 🧠 Musk vs. Memphis smog, AI that reads your mind, Cloudflare makes bots pay.

• 💡 Skim the coming rules now—future-you won’t be sprinting to catch up.

Let’s dive in. No floaties are needed…

The Laboratory

Why the EU’s AI Act delay could help big tech

Regulating a new technology, especially one that is rapidly evolving, has never been easy. Regulators may have dropped the ball on regulating social media and online content, but with AI they are taking a proactive rather than a reactive approach. A fine example of how governments view the transformational strength of artificial intelligence models is the EU’s AI Act.

The economic bloc has chosen a proactive approach to regulating foundational AI models to ensure transparency in their technical documentation, implementation, and compliance with the EU’s copyright laws. However, as is the case with regulations that can curb the powers of big corporations, increase costs, or stifle the rate of progress, the AI Act has been met with resistance.

Recently, the EU announced that a code of practice intended to help companies comply with its landmark artificial intelligence rules might only come into effect at the end of 2025, providing companies additional preparation time. The announcement comes even as big names in the industry, including Alphabet's Google, Meta Platforms, and European companies such as Mistral and ASML, have called for a delay in implementing the Artificial Intelligence Act, partly due to the lack of a code of practice. Several governments have also supported the calls.

So, what are the requirements under the EU’s AI Act, how could they set the tone for the rest of the world, and when exactly will they come into effect? Let us take a closer look.

What is the EU AI Act?

It is considered the world’s first comprehensive regulatory framework for artificial intelligence. The act takes a three-pronged approach to regulating AI; it prohibits some AI uses outright while setting standards for developing and deploying certain high-risk AI systems, and rules for general-purpose AI (GPAI) models.

The Act applies to multiple operators in the AI value chain, such as providers, deployers, importers, distributors, and product manufacturers. The Act also has provisions that bring providers outside the EU that offer AI services within the bloc to designated representatives to ensure coordinated compliance efforts.

The Act also lays down provisions for penalties for noncompliance. Organizations can be fined up to EUR 35,000,000 or 7% of worldwide annual turnover, whichever is higher. Penalties can also be imposed for the supply of incorrect, incomplete, or misleading information to authorities, and noncompliance with the requirements for high-risk AI systems. The Act, however, has different rules for fining start-ups and other small and medium-sized organizations. For these businesses, the fine is the lower of the two possible amounts as opposed to the ones for larger organizations.

What is the timeline for the Act to take effect?

The EU Act was passed in August 2024; however, it will be implemented in a staggered approach, with different provisions of the law going into effect in stages.

In the first stage, from 2 February 2025, the bloc implemented the rules laid down on prohibited AI practices.

In the second stage, from 2 August 2025, the rules for general-purpose AI would take effect for new GPAI models. Providers of GPAI models that were placed on the market before 2 August 2025 will have until 2 August 2027 to comply. From 2 August 2026, the rules for high-risk AI systems will take effect. From 2027, the rules for AI systems that are products or safety components of products regulated under specific EU laws will apply.

Why are AI companies seeking a delay?

The provisions that come into effect on August 2, 2025, apply to foundation models like those made by Google, Mistral, and OpenAI. These provisions require them to be subject to transparency requirements such as drawing up technical documentation, complying with EU copyright law, and providing detailed summaries about the content used for algorithm training. Companies are also required to test for bias and robustness under these rules.

As such, for AI companies, the enforcement of the act means additional costs for compliance. And for those who make AI models, the requirements are tougher.

Companies express uncertainty regarding compliance due to the absence of guidelines and have requested a two-year pause on the AI Act before key obligations take effect. Companies have also requested a simplification of the rules, claiming they could stifle innovation, particularly in Europe, where companies have smaller compliance teams than their U.S. counterparts.

How has the EU responded?

In response to calls for a delay in implementation, the EU published a Code of Practice for Large Language Models (GPAI) in July 2025. The code applies to models such as OpenAI's ChatGPT and similar models launched by Google and Mistral and was originally planned for May 2.

The Commission plans to present the code in the coming days and expects companies to sign up next month, and the guidance is likely to kick in at the end of the year, a report from Reuters said. While companies can refuse to sign the code, which is voluntary, those who decline will not benefit from the legal certainty provided to signatories.

Publishing the code indicates that the EU is resisting industry pressure to delay AI rules.

Could delays mean a strategic advantage for big tech?

Bram Vranken, Corporate Europe Observatory researcher and campaigner, told Reuters, "Delay. Pause. Deregulate. That is big tech's lobby playbook to fatally weaken rules that should protect us from biased and unfair AI systems." And he may be right.

The delayed rollout of the EU’s Code of Practice for general-purpose AI (GPAI) models is not just a procedural hiccup. It has the potential to significantly shift the balance of power in favor of large, U.S.-based technology firms such as Google, OpenAI, and Meta. While the EU’s intention behind the delay is to provide companies with clarity and breathing room, the practical outcome may be the consolidation of dominance by firms already far ahead in developing and deploying AI systems.

The postponement effectively gives tech giants more time to scale their AI offerings, expand user bases, and integrate foundational models into their product ecosystems without regulatory friction. This would allow them to cement their tools as industry standards before competitors are able to catch up or adapt to upcoming compliance obligations.

For companies that already enjoy data advantages, strong computing infrastructure, and global reach, even a few extra months can translate into significant headway.

Delays in implementation could also harm smaller startups, as they may find themselves devoting valuable time and capital to prepare for regulations. Meanwhile, U.S.-based companies have already had a head start due to massive venture funding, early adoption, and looser regulatory frameworks at home.

A risky move for Europe

The delay in implementing the EU’s Code of Practice for AI may appear to be a pragmatic move, intended to give the industry time to adapt.

However, it may unintentionally tilt the playing field further toward already dominant big Tech firms. By providing regulatory breathing room without immediate accountability, the EU may inadvertently strengthen the competitive edge of major U.S. companies that are already years ahead in developing and deploying foundation models.

This could cost European startups and smaller players, who are more likely to struggle with compliance and resource constraints.

While the EU remains committed to setting a global precedent with the AI Act, its cautious approach may undermine its very goal: reining in the unchecked power of large AI firms and ensuring fairness, transparency, and safety in artificial intelligence systems.

Bite-Sized Brains

• Musk’s Memphis Mess — xAI admits its data center plan could dump 1.7M tons of CO₂ a year near Memphis, sparking local outrage and fresh calls for greener AI.

• NYT: Does ChatGPT “think”? — Cognitive scientists find people still ascribe human minds to large-language models, even after being told they’re just math, raising new ethics headaches.

• Cloudflare to AI scrapers: pay up — In a new podcast, Cloudflare details its marketplace that lets any site charge Gemini, ChatGPT & pals every time they crawl your pages.

Roko’s Pro Tip

Monday Poll

Meme of the Day

Rate this edition