In the fast-evolving world of AI, open-sourcing models has become a battleground for innovation, ethics, and regulation. Just recently, on August 25, 2025 (yes, that’s today!), Elon Musk announced that xAI has open-sourced Grok 2.5, its flagship model from last year, making the weights available on Hugging Face. This move echoes OpenAI’s earlier release on August 5, 2025, of two open-source models: gpt-oss-120b (120 billion parameters) and gpt-oss-20b (20 billion parameters), under the permissive Apache 2.0 license with an added usage policy. Both companies are pushing the boundaries of general-purpose AI (GPAI) models — those versatile systems capable of tackling reasoning, coding, math, and more. But with great power comes great scrutiny, especially under the EU AI Act (Regulation (EU) 2024/1689), which sets strict rules for transparency, risk management, and open-source claims. Inspired by a recent Medium article analyzing OpenAI’s models (check it out here), I’ll conduct a similar compliance check for Grok 2.5. Using publicly available info like model cards and announcements, we’ll evaluate its alignment with the Act’s requirements for open-source GPAI models. Spoiler: It’s not as straightforward as it seems. Note that this is a high-level analysis — true compliance needs official regulatory review.Grok Breaking Down the Models: Grok 2.5 vs. OpenAI’s Duo Let’s start with the basics to set the stage. Grok 2.5 (xAI): This beast clocks in at around 270 billion parameters, trained back in 2024 on text-based tasks like reasoning. What’s released? The model weights (a hefty ~500 GB across 42 files) and the tokenizer. But details on the full architecture, training code, or datasets? Slim to none — only hints like a June 2024 knowledge cutoff. The license is a custom “Grok 2 Community License Agreement”: revocable, allows commercial and non-commercial use, but slaps on restrictions like banning its use to train or improve other AI models. No separate usage policy, just the license terms. gpt-oss-120b and gpt-oss-20b (OpenAI): Smaller siblings at 120B and 20B parameters, trained on trillions of filtered tokens with chain-of-thought tweaks. Releases include weights, architecture details, tokenizer (via TikToken), and even some training code snippets. Licensed under Apache 2.0 — super permissive — with a usage policy encouraging responsible AI without heavy restrictions. Training compute (measured in FLOPs) isn’t explicitly shared for Grok 2.5, but given its size, it’s probably under the 10²⁵ FLOPs mark that triggers “systemic risk” status — similar to GPT-3’s estimates (around 3.14 x 10²³ FLOPs). OpenAI’s models are in the same boat, as noted in the original article. The EU AI Act: What Open-Source GPAI Models Need to Nail The EU AI Act classifies GPAI as AI that handles diverse tasks without a narrow focus (Article 3, point 63). For “open-source” ones (Recital 102, Article 3 point 12), the bar is high: They must use a “free and open license” allowing unrestricted access, use, study, modification, and sharing — including derivatives — with no commercial bans or field limits. Key obligations include: Article 53 (Baseline for All GPAI Providers): Technical docs on training/testing, copyright respect (like honoring opt-outs from Directive 2019/790), and usage info. Article 55 (For Systemic Risks): If over 10²⁵ FLOPs or deemed high-risk, add risk assessments, testing, and reporting. Open-source models can snag exemptions from some transparency rules if their license is truly open and they’re not monetized (e.g., no subs). Exemptions and the Code of Practice: Genuine open-source (non-systemic) skips some hurdles if the license is barrier-free. The EU’s upcoming Code of Practice (rolling out in 2025) offers a voluntary roadmap for safety, transparency, and copyright. The Act loves open-source for sparking innovation but calls out licenses with sneaky restrictions. The Compliance Breakdown: Where Grok 2.5 Stands vs. OpenAI Mirroring the original article’s style, here’s a table assessing compliance based on public data. Ratings: “Likely Compliant” (checks out), “Partial/Questionable” (iffy spots), or “Potential Non-Compliance” (red flags). I’ve included OpenAI for direct comparison.https://medium.com/media/8218eea850b241bff8bd2a4f09d44233/href Wrapping It Up: Lessons for xAI and the AI World Grok 2.5 ticks some boxes as a GPAI release but stumbles on open-source purity thanks to its custom license’s revocability and restrictions — potentially stripping away exemptions and inviting deeper EU scrutiny under Articles 53 and 55. OpenAI’s gpt-oss models, with their straightforward Apache 2.0 setup and better docs, seem to sail through more smoothly, qualifying for those sweet exemptions while hitting baselines. If Grok 2.5’s FLOPs secretly top 10²⁵ (doubtful, but possible), the gaps widen. xAI could level up by switching to a standard open license and beefing up transparency. For anyone in AI, this highlights the Act’s push: Open-source is great, but only if it’s truly open. Curious about the EU’s full guidelines? Dive into them or chat with regulators for the real deal. What do you think — will more companies follow suit, or tighten up? Drop your thoughts below! xAI’s Grok 2.5: Open-Sourced, But Does It Pass the EU AI Act Test? was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this storyIn the fast-evolving world of AI, open-sourcing models has become a battleground for innovation, ethics, and regulation. Just recently, on August 25, 2025 (yes, that’s today!), Elon Musk announced that xAI has open-sourced Grok 2.5, its flagship model from last year, making the weights available on Hugging Face. This move echoes OpenAI’s earlier release on August 5, 2025, of two open-source models: gpt-oss-120b (120 billion parameters) and gpt-oss-20b (20 billion parameters), under the permissive Apache 2.0 license with an added usage policy. Both companies are pushing the boundaries of general-purpose AI (GPAI) models — those versatile systems capable of tackling reasoning, coding, math, and more. But with great power comes great scrutiny, especially under the EU AI Act (Regulation (EU) 2024/1689), which sets strict rules for transparency, risk management, and open-source claims. Inspired by a recent Medium article analyzing OpenAI’s models (check it out here), I’ll conduct a similar compliance check for Grok 2.5. Using publicly available info like model cards and announcements, we’ll evaluate its alignment with the Act’s requirements for open-source GPAI models. Spoiler: It’s not as straightforward as it seems. Note that this is a high-level analysis — true compliance needs official regulatory review.Grok Breaking Down the Models: Grok 2.5 vs. OpenAI’s Duo Let’s start with the basics to set the stage. Grok 2.5 (xAI): This beast clocks in at around 270 billion parameters, trained back in 2024 on text-based tasks like reasoning. What’s released? The model weights (a hefty ~500 GB across 42 files) and the tokenizer. But details on the full architecture, training code, or datasets? Slim to none — only hints like a June 2024 knowledge cutoff. The license is a custom “Grok 2 Community License Agreement”: revocable, allows commercial and non-commercial use, but slaps on restrictions like banning its use to train or improve other AI models. No separate usage policy, just the license terms. gpt-oss-120b and gpt-oss-20b (OpenAI): Smaller siblings at 120B and 20B parameters, trained on trillions of filtered tokens with chain-of-thought tweaks. Releases include weights, architecture details, tokenizer (via TikToken), and even some training code snippets. Licensed under Apache 2.0 — super permissive — with a usage policy encouraging responsible AI without heavy restrictions. Training compute (measured in FLOPs) isn’t explicitly shared for Grok 2.5, but given its size, it’s probably under the 10²⁵ FLOPs mark that triggers “systemic risk” status — similar to GPT-3’s estimates (around 3.14 x 10²³ FLOPs). OpenAI’s models are in the same boat, as noted in the original article. The EU AI Act: What Open-Source GPAI Models Need to Nail The EU AI Act classifies GPAI as AI that handles diverse tasks without a narrow focus (Article 3, point 63). For “open-source” ones (Recital 102, Article 3 point 12), the bar is high: They must use a “free and open license” allowing unrestricted access, use, study, modification, and sharing — including derivatives — with no commercial bans or field limits. Key obligations include: Article 53 (Baseline for All GPAI Providers): Technical docs on training/testing, copyright respect (like honoring opt-outs from Directive 2019/790), and usage info. Article 55 (For Systemic Risks): If over 10²⁵ FLOPs or deemed high-risk, add risk assessments, testing, and reporting. Open-source models can snag exemptions from some transparency rules if their license is truly open and they’re not monetized (e.g., no subs). Exemptions and the Code of Practice: Genuine open-source (non-systemic) skips some hurdles if the license is barrier-free. The EU’s upcoming Code of Practice (rolling out in 2025) offers a voluntary roadmap for safety, transparency, and copyright. The Act loves open-source for sparking innovation but calls out licenses with sneaky restrictions. The Compliance Breakdown: Where Grok 2.5 Stands vs. OpenAI Mirroring the original article’s style, here’s a table assessing compliance based on public data. Ratings: “Likely Compliant” (checks out), “Partial/Questionable” (iffy spots), or “Potential Non-Compliance” (red flags). I’ve included OpenAI for direct comparison.https://medium.com/media/8218eea850b241bff8bd2a4f09d44233/href Wrapping It Up: Lessons for xAI and the AI World Grok 2.5 ticks some boxes as a GPAI release but stumbles on open-source purity thanks to its custom license’s revocability and restrictions — potentially stripping away exemptions and inviting deeper EU scrutiny under Articles 53 and 55. OpenAI’s gpt-oss models, with their straightforward Apache 2.0 setup and better docs, seem to sail through more smoothly, qualifying for those sweet exemptions while hitting baselines. If Grok 2.5’s FLOPs secretly top 10²⁵ (doubtful, but possible), the gaps widen. xAI could level up by switching to a standard open license and beefing up transparency. For anyone in AI, this highlights the Act’s push: Open-source is great, but only if it’s truly open. Curious about the EU’s full guidelines? Dive into them or chat with regulators for the real deal. What do you think — will more companies follow suit, or tighten up? Drop your thoughts below! xAI’s Grok 2.5: Open-Sourced, But Does It Pass the EU AI Act Test? was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story