{"slug": "the-gpt-5-6-rollout-is-another-example-of-why-european-businesses-cant-rely-on", "title": "The GPT-5.6 rollout is another example of why European businesses can’t rely on", "summary": "OpenAI's GPT-5.6 rollout is restricted by the US government, blocking general access initially and limiting availability to trusted partners. This follows similar restrictions on Anthropic's models, highlighting a trend where European businesses are increasingly unable to rely on US AI technology due to national security controls.", "body_md": "Washington is now gating access to the most powerful new AI models coming out of US tech companies. Unlike its previous rollouts, OpenAI won’t be able to release GPT-5.6 to the general public initially.\n\nThis isn’t the first time an AI product has been blocked from general availability. Just earlier this month, Anthropic faced [restrictions(new window)](https://www.reuters.com/technology/us-blocks-foreign-access-anthropics-most-advanced-ai-models-axios-reports-2026-06-13/) on its new Fable 5 and Mythos models, with the EU even having to appeal to the US administration to gain access.\n\nAs new technology emerges in the US, European businesses are being left behind, and it’s accelerating with every new technological advance. Right now, this means AI models. But in the future, it could go further: blocking software updates, or even refusing to export hardware.\n\nThe US government is deciding who gets to use American technology, and in doing so they’re deciding Europe’s future.\n\n## What is GPT-5.6?\n\nGPT-5.6 actually consists of three versions: Sol, Terra, and Luna. Each one targets a different audience, with Sol (the flagship model) offering agentic capabilities, Terra being a more balanced model for everyday work, and Luna offering speed and affordability (at $1 input / $6 output, compared to Sol’s $5 input / $30 output.)\n\nOpenAI claims Sol is its strongest model, with an ‘ultra’ mode that “goes beyond the capabilities of a single agent by leveraging subagents to accelerate complex work.”\n\nIn recent months, more and more businesses have been adopting agentic systems. Autonomous [AI agents](/blog/ai-agent) work differently than the chatbots most people are familiar with — they combine the natural language processing capabilities of [large language models](/blog/llm) (LLMs) with the ability to access tools and memory to take action on their own.\n\nThis presents a new world of possibilities for businesses: Smaller businesses with limited resources can outsource simple administrative tasks, data analysis, infrastructure inspection, risk monitoring, or customer service. Agents can be embedded into existing workflows to increase efficiency and productivity for businesses of any size, which is exactly why it’s worrying that access to new agentic models could be cut off from the EU.\n\nIf this trend continues, European businesses will be left behind, unable to compete against their global peers because of their dependence on US tech.\n\n## Why the US is controlling the GPT-5.6 release\n\nIn its [release announcement(new window)](https://openai.com/index/previewing-gpt-5-6-sol/), OpenAI laid out how its new model will be released:\n\n“As part of our ongoing engagement with the US government, we previewed our plans and the models’ capabilities ahead of today’s launch. At their request, we are starting with a limited preview for a small group of trusted partners whose participation has been shared with the government, before releasing more broadly. During this preview, we will continue testing and coordinating closely with partners as we work toward broader availability.”\n\nConcerns have been raised in Washington about the potential national security risks created by new and powerful AI models. While OpenAI claims that “GPT‑5.6 is trained to refuse prohibited cyber assistance, including when users attempt to disguise their intent or jailbreak the model”, it also notes that “no single safeguard is sufficient against determined or adaptive misuse”. Because this new model will be capable of making autonomous real-world actions, it could be a more powerful tool for legitimate and criminal means.\n\nPresident Trump signed an [executive order(new window)](https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/) on June 2 ordering that a new classified benchmarking process for AI models be created by August 2026. The US government will use this process to “assess the advanced cyber capabilities of AI models and determine the threshold at which an AI model should be designated a ‘covered frontier model’”. AI developers must now provide the government with access to frontier models for a period of up to 30 days before they plan to release them.\n\nWhile OpenAI said they “don’t believe this kind of government access process should become the long-term default”, there’s no public evidence to justify their optimism. OpenAI itself is even working with the government to develop the cyber framework, meaning it’ll be able to lobby for a framework that best suits Big Tech’s interests.\n\n## Why should European businesses be concerned?\n\nThe laws, executive orders, and national security frameworks shaping US tech were written with American interests in mind. European businesses are an afterthought at best, and a controlled market at worst.\n\nAs a business leader in Europe, the US restrictions on AI raise several red flags. In the future, you could suddenly lose access to existing software (known as a kill switch) or miss out on new innovation. As a result, your US competition could gain the upper hand. And even when you retain access to software, the vendor terms you relied on could suddenly change, as Copilot customers learned the hard way when [Microsoft introduced flex routing](/business/blog/microsoft365-copilot-flex-routing).\n\nThe US government is willing to shut Europe and the rest of the world out of the most advanced tech available. The solution won’t come from America. It will come from European investment in a sovereign tech stack.\n\n## It’s time to choose European tech\n\nIf the US can gate access to new models, it can restrict access to the services European businesses are already running on. All it will take is an executive order, an export control decision, or a shift in geopolitical relations.\n\nAlongside America’s increasing national [tech protectionism](/business/blog/us-tech-risk-report-for-europe), the [European tech sovereignty](/business/europe-tech-watch) movement has grown. Businesses have woken up to the reality that relying on US tech means losing control of your data and the reality that the US could switch off your access. Ironically, the US Ambassador to the EU recently warned that the [European Chips Act](/business/blog/eu-tech-sovereignty-package) “doesn’t sound very consistent with the EU-US trade framework agreement.”\n\nTo help address this challenge, we launched [Proton Workspace](/business/plans) this year, giving any business — including American firms — the ability to reduce their overreliance on Big Tech. It’s a secure alternative that gives organizations email, cloud storage, VPN, AI assistant, video conferencing, and more, without the surveillance and US government overreach.\n\nBy design, Proton is:\n\n- Private and encrypted by default\n- Open source and audited by third party security experts\n- Built with compliance in mind\n- Sovereign and protected from US surveillance\n\nA European business suite doesn’t just break dependence on US tech and protect businesses from being shut out of their tools. It’s also a way to actively invest in an independent European tech sector that prioritizes business data protection and growing the European economy. If you want a world where access to the most advanced tech isn’t decided by your geographic location, consider moving away from US tech companies; they’re certainly moving away from you.", "url": "https://wpnews.pro/news/the-gpt-5-6-rollout-is-another-example-of-why-european-businesses-cant-rely-on", "canonical_source": "https://proton.me/business/blog/openai-gpt-5-6", "published_at": "2026-06-29 16:35:32.422819+00:00", "updated_at": "2026-06-29 16:35:34.434896+00:00", "lang": "en", "topics": ["large-language-models", "ai-policy", "ai-safety", "ai-agents"], "entities": ["OpenAI", "Anthropic", "GPT-5.6", "Fable 5", "Mythos", "Sol", "Terra", "Luna"], "alternates": {"html": "https://wpnews.pro/news/the-gpt-5-6-rollout-is-another-example-of-why-european-businesses-cant-rely-on", "markdown": "https://wpnews.pro/news/the-gpt-5-6-rollout-is-another-example-of-why-european-businesses-cant-rely-on.md", "text": "https://wpnews.pro/news/the-gpt-5-6-rollout-is-another-example-of-why-european-businesses-cant-rely-on.txt", "jsonld": "https://wpnews.pro/news/the-gpt-5-6-rollout-is-another-example-of-why-european-businesses-cant-rely-on.jsonld"}}