人工智能正从人们“提问”的对象，转变为人们“投入使用”的工具。
OpenAI 首次发布按国家划分的数据，展示世界各地的人们如何使用 ChatGPT。这些发现揭示了一项技术正超越“提供答案”，进入“采取行动”的阶段，正从早期采用者扩展到更广泛的人群，并以更多方式为更多人带来价值：
- 从提问到行动： 在工作中，人们使用 ChatGPT 完成任务或进行创作（从写作、编程到分析）的可能性，是工作之外的两倍多。
- AI 采用正变得更加全球化： 拉丁美洲、非洲和大洋洲的国家正在追赶早期采用者，缩小全球采用差距。
- AI 正超越文本： 多媒体是全球增长最快的用例，占消息总量的 7.8%，在巴西和哥伦比亚等国家，这一比例超过十分之一。
- AI 正触及早期采用者之外的人群： 在几乎所有国家，35 岁以上人群的使用率都在上升。在法国和捷克，他们发送的消息占比在过去一年中上升了超过 10 个百分点。
综合来看，这些数据指向一个明显的转变：AI 不再仅仅是帮助人们找到答案。它正在帮助更多地方、更多的人完成事务。
这些新数据可在 OpenAI Signals 上查看或下载，这是我们 OpenAI 经济研究团队的数据、研究和分析中心。这些国家级统计数据将帮助包括世界各地的政策制定者和研究人员在内的人们，更好地了解超过 10 亿人如何将 ChatGPT 投入使用。
Signals 数据集具体反映了在 ChatGPT Free、Go、Plus 和 Pro 账户内发送的消息——这些账户通常由个人而非组织管理。
在工作中，“行动”占主导
数据反映了一个近期头条新闻中出现的趋势，尤其是在 ChatGPT Work 推出之后：人们越来越多地使用这项技术，特别是用于专业用途。在这些场景中，用户更频繁地“行动”——使用 ChatGPT 产生输出或执行任务。
这类工作包括编辑、编码或进行分析等应用。我们的数据显示，在全球范围内，人们在工作时以这种方式使用 ChatGPT 的可能性是工作之外的两倍多。在工作之外，使用更具探索性：“提问”，即寻求信息和澄清，仍然是最大的类别。
采用差距正在缩小
延续 6 月 记录的全球稳步攀升趋势，那些以较低人均 ChatGPT 采用率开始 AI 旅程的国家和地区，正逐渐赶上已确立的早期采用者。
对于 2026 年第二季度，我们的团队更新了各国使用量排名（参见此前的季度排名 此处）。下图显示了自 2026 年第一季度以来人均排名的变化。
值得注意的是，在第二季度，拉丁美洲、大洋洲和非洲部分地区的使用量增长速度超过了世界其他地区，其中秘鲁、乌拉圭和哥斯达黎加在全球排名中上升最多。北美和欧洲的采用率持续上升，但南半球部分地区的人均采用率正在迎头赶上。
国家人均消息数排名变化，2026 年第一季度至 2026 年第二季度（正数 = 提升）
注意：对于 144 个国家，蓝色表示排名提升，棕色表示下降，灰色表示无变化。阴影线标识被排除或不符合条件的国家。
随着新功能的出现，特定用例会吸引人们的注意力并推动采用。以多媒体生成、分析和检索为例。自 2026 年 4 月 ChatGPT Images 2.0 发布以来，全球专注于多媒体使用的消息占比上升至 7.8%。虽然它仍落后于实用指导、写作和信息寻求等主要用例，但多媒体使用量今年迄今一直持续上升。
深入分析多媒体消息类别，数据显示出国家和地区之间的显著差异。下图展示了拉丁美洲的增长，在巴西和哥伦比亚等国家，超过十分之一的 ChatGPT 消息被归类为多媒体。
注意：2026 年第二季度，按国家划分，被归类为多媒体的已分类个人 ChatGPT 消息占比。本次更新发布包含 126 个具有合格第二季度多媒体估算值的国家；条纹国家不在发布范围内或没有合格数据。
全球 35 岁以上人群的 ChatGPT 采用率增长
人均采用率的环比变化不仅限于用例和地区。数据还显示了不同年龄段人群的采用差异。
过去一年的趋势线显示，在几乎所有国家，35 岁以上人群发送的消息占比都有所增加。同比来看，这些用户现在发送的消息占比比 12 个月前高出 5 个百分点。
这仅是对在 ChatGPT 平台上自我报告年龄的用户进行的分析。
逐国查看数据，35 岁及以上人群的采用率在多个国家增长尤为迅速。例如，下图显示了几个欧洲国家中，35 岁及以上用户发送的 ChatGPT 消息占比的显著变化。在法国和捷克，35 岁及以上用户发送的消息占比在过去一年中上升了超过 10 个百分点。
近四分之三的欧洲国家在这些用户发送的消息占比上出现了高于平均水平的增长，而一些东南亚国家（如新加坡）则出现了相反的模式。总体而言，该地区八个国家中有六个国家的这些用户发送的消息占比有所增加，尽管增幅很小。
35 岁以上用户发送的按年龄分类消息的三个月移动平均占比，相对于各国 2025 年第二季度平均值（百分点）
注意：35 岁及以上用户发送的自我报告、按年龄分类的个人消息的三个月移动平均占比；变化相对于各国自身的 2025 年第二季度平均值。灰色线条显示所有 111 个具有完整估算值的国家；灰色虚线显示各国中位数。2026 年 6 月端点将最新季度与 2025 年第二季度进行比较。来源：OpenAI Signals 更新的公开发布 CSV 文件。
关于 OpenAI Signals 的更多信息
OpenAI Signals是OpenAI经济研究团队面向全球分享数据和研究的平台。除了我们的研究，我们还定期发布关于个人和企业如何使用ChatGPT的最新动态。我们鼓励您在此处下载数据，并在此处（在新窗口中打开）探索我们的方法论。
AI is moving from something people ask to something they put to work.
For the first time, OpenAI is publishing country-by-country data showing how people around the world use ChatGPT. The findings reveal a technology that is moving beyond answers and into action, spreading beyond early adopters, and becoming useful to more people in more ways:
- From asking to doing: At work, people are more than twice as likely to use ChatGPT to complete a task or create something, from writing and coding to analysis, than they are outside work.
- AI adoption is becoming more global: Countries across Latin America, Africa, and Oceania are catching up to early adopters, narrowing the global adoption gap.
- AI is moving beyond text: Multimedia is the fastest-growing use case globally, accounting for 7.8% of messages and more than one in ten in countries including Brazil and Colombia.
- AI is reaching beyond early adopters: Usage among people over 35 is rising in nearly every country. In France and Czechia, their share of messages increased by more than 10 percentage points in the past year.
Taken together, the data points to a clear shift: AI is no longer just helping people find answers. It is helping more people, in more places, get things done.
This new data is available to view or download on OpenAI Signals, our hub for data, research, and analysis, from the OpenAI Economic Research Team. These country-level statistics will help people, including policymakers and researchers around the world, better understand how more than 1 billion people are putting ChatGPT to work.
The Signals dataset specifically reflects messages sent within ChatGPT Free, Go, Plus, and Pro accounts—the group of accounts generally managed by individuals rather than organizations.
At work, “doing” dominates
The data reflect a trend that has appeared in recent headlines, especially since the rollout of ChatGPT Work: people are increasingly using the technology, particularly for professional pursuits. In these settings, users are more frequently “doing”—using ChatGPT to produce an output or perform a task.
This type of work includes applications such as editing, coding, or conducting an analysis. Our data shows that, across the globe, people are more than twice as likely to use ChatGPT this way at work than outside of work. Outside of work, use is more exploratory: “asking,” or seeking information and clarification, remains the largest category.
The adoption gap is closing
Continuing the steadily climbing trend worldwide documented in June, the countries and regions that began their AI journeys with lower ChatGPT per-capita adoption rates are gradually catching up to the established early adopters.
For Q2 2026, our team updated our usage rankings across countries (see previous quarterly rankings here). The chart below shows how the per-capita leaderboard changed since the first quarter of 2026.
Notably, in the second quarter, usage in parts of Latin America, Oceania, and Africa increased faster than usage in other parts of the world, with Peru, Uruguay, and Costa Rica rising the most among countries in global rankings. Adoption continues to rise in North America and Europe, but parts of the Southern Hemisphere are catching up in per-capita adoption rates.
Change in country messages-per-capita rank, Q1 2026 to Q2 2026 (positive = improved)
Note: For 144 countries, blue indicates an improvement in rank, brown a decline, and gray no change. Hatching identifies excluded or ineligible countries.
As new capabilities emerge, specific use cases command people’s attention and drive adoption. Take multimedia generation, analysis, and retrieval, for example. Since the release of ChatGPT Images 2.0 in April 2026, the share of messages across the world focused on multimedia use increased to 7.8%. While it still lags behind the leading use cases of practical guidance, writing, and information-seeking, multimedia use has been on a consistent upswing year-to-date.
Drilling deeper into the category of multimedia messages, the data shows distinctive differences among countries and regions. The graphic below demonstrates the uptick in Latin America where, in countries like Brazil and Colombia, more than one in ten messages to ChatGPT is categorized as multimedia.
Note: Share of classified individual ChatGPT messages categorized as multimedia, by country, in Q2 2026. The refreshed release includes 126 countries with an eligible Q2 multimedia estimate; striped countries are outside the release or have no eligible data.
ChatGPT adoption grew worldwide among people over 35
The quarter-over-quarter changes in per-capita adoption are not limited to use cases and regions alone. The data also demonstrate differences in adoption by age cohort.
The trend line over the last year shows that the percentage of messages sent by people over 35 increased in almost every country. Year over year, these users now account for a 5% higher share of messages than they did 12 months earlier.
This is an analysis of only users who self-reported their age on the ChatGPT platform.
Looking at the data country-by-country, adoption among people aged 35 and older is growing particularly quickly in several countries. The chart below, for example, shows how the share of ChatGPT messages sent by users 35 and above shifted significantly in several European countries. In France and Czechia, the share of messages sent by users 35 or older increased by more than 10 percentage points in the last year.
Almost three-quarters of European countries saw larger-than-average increases in their share of messages sent by these users, while the opposite pattern emerged in some Southeast Asian countries, such as Singapore. Overall, the share of messages sent by these users increased in six of the eight countries in the region, albeit by only a small margin.
Three-month trailing share of age-classified messages from users aged 35+, relative to each country’s Q2 2025 average (percentage points)
Note: Three-month trailing share of self-reported, age-classified individual messages from users aged 35 and older; changes are relative to each country’s own Q2 2025 average. Gray lines show all 111 countries with complete estimates; dashed gray shows the equal-country median. June 2026 endpoints compare the latest quarter with Q2 2025. Source: OpenAI Signals refreshed public-release CSVs.
More about OpenAI Signals
OpenAI Signals is the OpenAI Economic Research team’s platform for sharing our data and research with the world. In addition to our research, we publish regular updates on how individuals and businesses use ChatGPT. We encourage you to download the data here and explore our methodology here(opens in a new window).
本文内容采集自官方网站，排版和翻译可能与原页面存在差异。
阅读官方全文