Countries where ChatGPT is banned in 2026: full list
Moveo AI Team
in
🤖 AI automation
In March 2023, Italy's Garante per la Protezione dei Dati Personali became the first Western regulatory authority to block ChatGPT, citing violations of the European General Data Protection Regulation.
The block lasted less than a month. In late 2024, the same authority fined OpenAI €15 million for inappropriate use of personal data in model training. In March 2026, the Court of Rome annulled that fine.
Between the first block and the fine annulment, a growing list of countries moved to restrict access to ChatGPT for reasons ranging from data sovereignty to political control of information.
This guide maps the countries where ChatGPT is blocked, restricted, or simply unavailable in 2026, with a primary source for each case and the current status of each restriction.
It covers three distinct categories that often get confused in similar lists: explicit government bans, OpenAI-side restrictions for legal or commercial reasons, and regulatory pauses that lasted shorter than expected.
At the end, the article covers operational implications for enterprises running customer service, accounts receivable, or other Customer-to-Cash operations across multiple jurisdictions.
The 2026 landscape: how many countries restrict ChatGPT?
In 2026, ChatGPT is blocked, restricted, or unavailable in approximately 20 countries, according to a survey by Cybernews referenced by Visual Capitalist. The aggregate number, however, hides an important distinction. Restrictions fall into three categories with different mechanics:
First Category
The first category is explicit government bans. Countries like China, Russia, North Korea, Iran, Cuba, and Syria block access to ChatGPT as part of broader information control regimes.
In these cases, the block comes from local government decisions, generally mediated by telecommunications authorities or digital security agencies. Russia's Roskomnadzor and China's CAC are the most well-known examples.
Second Category
The second category is OpenAI-side restriction. The company maintains a public list of supported countries and territories, and any access attempt from outside this list results in blocking.
Some of these exclusions overlap with government bans (Russia falls into both categories, for example), but others are unilateral OpenAI decisions motivated by international sanctions, legal complexity, or commercial unviability. Hong Kong and Belarus, for example, are cases where OpenAI chose not to operate even without an explicit ban from the local government.
Thrid Category
The third category is the regulatory pause. Italy pioneered this model in 2023 with a block that lasted just over four weeks, and since then data protection authorities in France, Spain, Germany, and Poland have opened similar investigations.
None resulted in prolonged blocks, but the "investigate first, decide later" pattern changed how OpenAI and its competitors operate in the European market. The Italian Garante remained active, blocking DeepSeek (Chinese competitor) in 2025 for refusing to cooperate with the European regulator.
Countries with full government bans
This category includes countries that block ChatGPT by local government decision, generally as part of broader internet control regimes.
The blocks are typically bilateral: the government blocks, and OpenAI also doesn't operate due to sanctions, local regulation, or commercial viability issues.
China
China has never officially allowed access to ChatGPT since the product's launch in November 2022. The block is part of the broader regime known as the Great Firewall, which filters Western services like Google, Facebook, Twitter, and WhatsApp.
The government justification for ChatGPT's exclusion combines three arguments: information control, digital sovereignty, and protection of the domestic AI market.
The Cyberspace Administration of China (CAC) maintains a mandatory registration regime for generative AI services operating in the country, requiring prior approval, compliance with local censorship, and data storage in Chinese territory. OpenAI never sought this registration.
The consequence is a mature domestic ecosystem of Chinese LLMs. Alibaba launched Qwen, Baidu launched ERNIE, ByteDance launched Doubao, Tencent launched Hunyuan, and the startup DeepSeek became an international reference in 2024 and 2025.
For companies operating in the Chinese market, the alternative is not use ChatGPT via VPN (a strategy that violates both OpenAI's terms and local regulation), but to adopt one of the approved local alternatives.
OpenAI, in turn, banned multiple accounts linked to Chinese entities in 2025 for misuse for surveillance and influence operations, according to its own security reports.
Russia
Russia is the most consulted case in international searches on the topic, partly because of its hybrid nature.
The block is simultaneously governmental and OpenAI-side. Roskomnadzor, the Russian communications agency, restricts access to a series of Western services, and OpenAI actively removed access from Russian territory starting in mid-2024, according to BankInfoSecurity reporting. The company sent communications to users informing that its infrastructure would begin blocking traffic from unsupported regions, including Russia, China, Iran, and Saudi Arabia.
As in China, the void produced a domestic ecosystem. Sberbank launched GigaChat as a Russian alternative to ChatGPT, and Yandex developed YandexGPT. Both operate within local regulation and store data in Russian territory, which for financial services and regulated sectors is the only viable option.
OpenAI has also maintained an active position in monitoring misuse by Russian actors. In 2025 and early 2026, the company published threat intelligence reports documenting the banning of hundreds of accounts linked to Russian influence operations, including the "Rybar" network and the development of the ScopeCreep trojan, as reported by The Hacker News and Axios.
For international companies with Russian operations, the scenario is clear: ChatGPT does not work, local alternatives are mandatory for internal use, and any AI architecture must contemplate data residency in Russian territory.
The official OpenAI list of supported countries is the canonical reference for confirming current status.
North Korea
North Korea maintains one of the most restrictive internet regimes in the world. For most citizens, access to the global internet is completely blocked, with access permitted only to a controlled domestic intranet called Kwangmyong.
ChatGPT, like any other public internet service, is out of reach for structural reasons rather than just regulatory ones.
The North Korean case differs from the others on this list because the ChatGPT block is not an AI regulatory decision, but a consequence of the country's general regime of informational isolation.
There is an additional dimension. OpenAI documented in its 2025 misuse report schemes linked to the North Korean IT worker program abroad, which uses accounts in third countries to generate fraudulent application materials for technology jobs at Western companies.
The company banned these accounts, but the operation confirms that actors linked to the regime get indirect access from intermediate jurisdictions.
Iran
Iran blocked ChatGPT shortly after its launch, as part of the country's broader internet control regime. The Iranian government maintains filters on a substantial number of Western services (Facebook, Twitter, YouTube, Instagram in various periods), and ChatGPT entered this list for two combined reasons.
The first is political concern: a language model capable of discussing sensitive topics in Farsi and producing content critical of the regime is seen as a potential mobilization tool.
The second is the set of US sanctions that make OpenAI's operation in the country unviable, even if the government allowed it.
OpenAI confirmed in its supported countries list that Iran is outside the operation's scope. As in previous cases, Iranian actors were identified in misuse operations.
OpenAI banned accounts linked to the Storm-2035 operation, which used ChatGPT to generate comments in English and Spanish promoting causes aligned with Iranian interests, as reported by The Hacker News in June 2025.
Cuba
Cuba also appears on the official list of countries where OpenAI does not operate, due to commercial sanctions imposed by the United States.
The practical issue is less about AI regulation and more about international trade: the US embargo prevents American companies from providing services to the Cuban government or residents without specific authorization from the Office of Foreign Assets Control. For OpenAI, the lowest-risk path is simply not to serve the country.
Internally, Cuba maintains a highly regulated internet system, with limited access to the global internet for most citizens and state-controlled providers (ETECSA). Even if OpenAI sought to serve the market, access would be filtered by local infrastructure.
For international companies with Cuban operations, this scenario implies that ChatGPT is not an available tool, and any AI strategy must contemplate this unavailability from the design phase.
Syria
Syria combines three factors that make ChatGPT unavailable: US sanctions that prevent OpenAI from operating in the country, an internal censorship regime that filters services considered politically risky, and a telecommunications infrastructure severely damaged by the civil war.
The combination produces a scenario in which, even if regulation and sanctions were relaxed, practical operation would be difficult.
The country appears both on the list of government bans and on the list of countries where OpenAI chose not to operate.
Other countries with restrictions
Beyond the six countries already covered, 2025 and 2026 surveys identify other jurisdictions where ChatGPT is inaccessible.
Afghanistan, Eritrea, South Sudan, Eswatini, Chad, Central African Republic, and Yemen appear as countries where, due to a combination of governmental instability, sanctions, limited internet infrastructure, or explicit control policy, the service is not available.
Yemen, in particular, adds to this list the complication of the ongoing civil war that limits the operation of any international service.
Belarus enters a particular category. The country does not explicitly block ChatGPT, but OpenAI chose not to serve the region, partly due to political alignment with Russia and related sanctions. The practical result for the user is the same, but the mechanism is different.
It's important to note that this list changes. Countries can come and go as government decisions, international sanctions, or adjustments in OpenAI's operation evolve. The most reliable reference, always, is the official list maintained by the company.\
Does your operation serve customers across multiple jurisdictions and you need to understand how country-level bans affect your conversational AI strategy?
Countries with partial restrictions or regulatory pauses
Not every country that has appeared on ChatGPT block lists maintains those restrictions.
The regulatory pauses category, which lasted just a few weeks in some cases, is important for understanding how the scenario evolved and what to expect from new investigations.
Italy: the emblematic case
Italy was the first Western country to block ChatGPT, on March 31, 2023. The Garante per la Protezione dei Dati Personali, the Italian data protection authority, ordered the suspension on three grounds: absence of legal basis for personal data processing in training, lack of transparency to users, and absence of age verification mechanisms. The block lasted approximately four weeks.
Timeline of the Italian case:
March 2023: ChatGPT blocked by the Garante.
April 2023: Service restored after OpenAI's corrective measures.
December 2024: €15 million fine issued to OpenAI for GDPR violations.
March 2025: First-instance decision suspended by the Court of Rome.
March 2026: Fine annulled by the Court of Rome.
The annulment of the fine in March 2026 ended, at least for now, what was the only final sanction applied in Europe against a generative AI company related to the model launch period. The Garante remains the most active authority in the European bloc on AI platform enforcement.
In 2025, it applied a similar block to DeepSeek, the Chinese competitor, for refusing to cooperate with the regulator. The Chinese service remains blocked in the Italian market.
Other European authorities that investigated
The Garante's 2023 move inspired equivalent authorities in other European Union countries to open their own investigations into ChatGPT.
Spain's AEPD, France's CNIL, Poland's UODO, and several German state authorities analyzed the service under GDPR. Most of these investigations are still formally ongoing or were transferred to the Irish Data Protection Commission after OpenAI established its European headquarters in Dublin in early 2024, which triggered GDPR's one-stop-shop mechanism.
None of these processes resulted in market blocks for ChatGPT, but ongoing regulatory pressure keeps the company's European operation under elevated scrutiny.
The consequence is that, although ChatGPT is available in all 27 European Union countries in 2026, the regulatory environment is the most active in the world.
The EU AI Act, which entered into force in phases throughout 2025 and 2026, added an additional layer of transparency and documentation obligations for general-purpose AI systems, classifying ChatGPT in a category subject to specific disclosure requirements.
Countries where OpenAI chose not to operate
The OpenAI does not operate category is distinct from government bans and deserves separate treatment.
In some countries, ChatGPT is inaccessible not because the local government banned it, but because OpenAI made the commercial or legal decision not to serve those regions.
The official supported countries page from OpenAI lists the served jurisdictions. Anything outside that list is, by default, unavailable.
The factors that lead OpenAI to not serve a jurisdiction vary. International sanctions (United States against Cuba, Syria, Iran, North Korea, regions of Crimea and Donetsk-Luhansk) eliminate the possibility of commercial operation. Local legal complexity (mandatory registration regimes like the Chinese, requirements for data storage in national territory, specific licensing requirements) also weighs in.
In other cases, the calculation is simpler: the market is too small to justify compliance investment. Hong Kong is the most cited example in this category, although status may change.
For globally operating companies, the practical implication is that a country's status on OpenAI's list can change without broad notice.
The official page should be consulted before any architecture decision that depends on ChatGPT in a specific jurisdiction.
Trying to bypass unavailability via VPN or other mechanisms violates the terms of service and can result in suspension or banning of the corporate account, according to OpenAI's explicit notice.
Will ChatGPT be banned in my country?
The question about future bans is probably the most important for companies defining AI strategy now.
The short answer is that the global trend in 2026 does not point toward broad bans of ChatGPT in Western markets, but rather toward increasingly detailed regulation defining how the service can be used in specific contexts.
The regulatory trajectory unfolds on three fronts.
European Union: the EU AI Act
The EU AI Act was approved in 2024 and entered into force in phases throughout 2025 and 2026. The legislation classifies AI systems by risk level (unacceptable, high, limited, minimal) and imposes proportional obligations.
For general-purpose generative AI systems like ChatGPT, the main obligations are transparency (informing the user that they are interacting with an AI), technical documentation, and supervision of models with systemic capability. The regulation is not prohibitive, but adds a compliance layer that affects how the service is offered.
Italy, through the Garante, remains the most active authority in the European bloc on enforcement of AI platforms. The €15 million fine on OpenAI in December 2024 (annulled in March 2026) and the DeepSeek block in 2025 signal that the Italian regulator does not hesitate to apply tough measures when it believes there has been a violation. Companies operating in European markets must contemplate this reality in their operational design.
United States: a sector-by-sector approach
The United States opted for a fragmented approach. Instead of comprehensive federal legislation, regulation unfolds through sectoral agencies.
The SEC and OCC published specific guides for AI use in financial services.
The Department of Health and Human Services (HHS) issued guidance for the healthcare sector under the HIPAA framework.
The Department of Defense established directives for AI use in military contexts.
Some state laws, such as the Colorado AI Act with effect expected in 2026, are beginning to impose specific obligations for high-risk systems.
The consequence is that a company operating in the United States needs to understand the regime applicable to its sector, not just a general regulation.
China: mandatory registration regime
China continues to apply the most restrictive regime among the major economies, with the CAC requiring prior registration of any generative AI system operating in the country, compliance with censorship rules, and data storage in Chinese territory.
This regime, instituted in 2023 and expanded in subsequent years, has kept ChatGPT out of the Chinese market since launch. For companies operating in China, this is no longer a hypothesis: it is a consolidated scenario, with domestic alternatives (Qwen, ERNIE, Doubao, DeepSeek) being the practical path.
Latin America: regulation in development
Brazil approved PL 2338/2023 (Marco Legal da IA) in its Senate in December 2024, with the bill currently moving through the House of Deputies. The vote, initially expected for late 2025, was postponed to February 2026.
The structure follows logic close to the EU AI Act: classification by risk level, proportional obligations, supervision shared between the data protection authority (ANPD) and sector regulators.
ChatGPT, Gemini, Claude, and similar are classified as low to moderate risk for most uses, with more intense regulation only in high-risk contexts such as HR, credit, healthcare, or justice.
Other Latin American countries (Mexico, Argentina, Chile, Colombia, Peru) are at different stages of legislative development. No Latin American country has implemented bans on ChatGPT, and existing restrictions are limited to specific uses in particular government bodies.
How country-level bans affect cross-border enterprises
For a company running across multiple jurisdictions (customer service, accounts receivable, collections), the fragmented set of bans and regulations creates a concrete operational challenge.
Three practical points deserve attention:
Service continuity
If a company depends on ChatGPT for some operational function and part of its customers or employees is in a country where the service is blocked, there is a gap that needs to be filled.
Companies that bet everything on a single public LLM platform without an enterprise alternative are exposed to unexpected interruptions (changes in the official list of supported countries, international sanctions, regulatory decisions).
For an operation that needs to serve customers 24/7 across multiple countries, this gap is unacceptable.
Data sovereignty
Even where ChatGPT is available, regulatory requirements vary. Europe's GDPR, Brazil's LGPD, the United States' HIPAA, and specific sectoral regulations impose restrictions on what data can be sent to servers outside the jurisdiction, under what conditions, and with what documentation.
A multi-country operation needs an architecture that allows configuring data destination by jurisdiction, not a single architecture that assumes all customers can be processed by the same pipeline.
Customer experience consistency
A financial services company offering AI agent customer service cannot have substantially different response quality between customers based in São Paulo, Mexico City, or Frankfurt simply because the underlying stack changes in each country.
The continuity of customer context, interaction history, and assumed commitments needs to function identically regardless of the jurisdiction in which the customer resides.
This is particularly critical in collections operations, where a promise made by the customer in one interaction needs to be respected in the next, even if the next occurs in a different channel or context.
The enterprise alternative: governance over geographic dependency
Companies that treat conversational AI as critical infrastructure rather than a point utility are migrating to enterprise platforms that offer the reasoning capabilities of LLMs without exposing the operation to the geographic uncertainty of public ChatGPT.
Moveo.AI's architecture was designed exactly for this scenario: Customer-to-Cash operations connecting customer service, accounts receivable, and collections across multiple jurisdictions, with distinct local regulatory requirements.
The platform is built around two central layers:
TrueThread is the persistent memory layer that captures every interaction, signal, and decision across the customer lifecycle, keeping that context within the company's own environment.
TruePath is the governed execution layer that applies policies, regulatory requirements, and approval structures to every automated action.
The result is a system in which the productivity benefits of conversational AI are decoupled from the geographic and regulatory restrictions of public ChatGPT.
For an operation serving customers in markets with different regulations, the advantage is direct.
The company defines where data is stored, under which regulation it operates, what actions the agent can execute in each jurisdiction, and maintains context continuity across interactions regardless of channel.
The agent that converses with the European customer in compliance with GDPR is the same system that converses with the Brazilian customer under LGPD and with the American customer under HIPAA, with context preserved in each case.
ChatGPT bans by country: what to consider in 2026
ChatGPT is blocked in approximately 20 countries in 2026, but the aggregate number matters less than the underlying trend.
AI regulation is becoming more detailed in almost all Western markets, and the architecture that worked in 2023 ("use public ChatGPT for everything") is becoming unviable for companies with serious operations in multiple jurisdictions.
The question for most companies in 2026 is not "will ChatGPT be banned where I operate?", but "how do I build an AI architecture that continues to function regardless of regulatory changes?".
Enterprise platforms like Moveo.AI offer the operational answer to that question. The combination of TrueThread and TruePath allows the company to maintain control over its data, its governance, and its customer context, while benefiting from the reasoning capabilities of language models.
This approach is what distinguishes an operation that depends on the global availability of ChatGPT from an operation that controls its own regulatory destiny.
*Last updated: April 2026. The list of countries where ChatGPT is blocked may change as government decisions, international sanctions, or adjustments to OpenAI's operation evolve. To confirm the current status of any specific country, consult the official list maintained by OpenAI.