ChatGPT turns 3: 800M weekly users and a remade world

ChatGPT just turned three, and the numbers are frankly staggering.

Launched on November 30, 2022, OpenAI’s chatbot has ballooned from a viral demo into an everyday fixture for roughly 800 million weekly active users across more than 20 languages. While born in Silicon Valley, its fastest-growing user bases are now in India, Brazil, Indonesia, and the Philippines.

Looking back at its first three years, the global footprint is wildly uneven, fascinating, and at times messy.

Courtrooms, farms, and robot dolls

Job disruption arrived fast. By 2023, freelance essay writers in Kenya, digital illustrators, translators, and web developers were already feeling the squeeze.

Yet workers across emerging markets quickly co-opted the tech. Indonesian filmmakers use ChatGPT for scriptwriting, Malawian farmers query localized bots for agricultural tips, and in Colombia, a staggering 85% of judges rely on free versions of ChatGPT or Copilot to draft rulings and clear backlogs.

In South Korea, senior citizens now chat with robot companion dolls powered by ChatGPT that dispense medication reminders and trigger emergency calls.

Education has seen both breakthroughs and blunders. In Mali, AI projects translate study materials into native dialects to keep kids in school. Meanwhile, South Korea pulled the plug on an official AI classroom initiative after only four months due to heavy pushback over hallucinations, privacy risks, and teacher burnout.

Then came politics. In 2024, with over 2 billion people voting worldwide, Indian political agencies raked in millions generating campaign collateral, while over 170 fake X accounts in Ghana used ChatGPT to seed political disinformation.

The language barrier and the sovereign AI push

While ChatGPT handles English, Spanish, and Japanese with ease, it falls apart in lower-resource languages like Bengali, Swahili, Urdu, and Thai. Tests reveal it often invents words, loses logic, or generates outright gibberish.

That linguistic divide has sparked a global wave of homegrown alternatives. The Philippines is developing ITanong to capture Taglish, Chile is spearheading Latam-GPT, and Nigeria tapped startup Awarri to build a state-backed model. Even founders in Mongolia are training local LLMs to escape Western API dependencies.

Then there is China, where OpenAI remains officially blocked. Users still bypass restrictions for everything from corporate slides to dating advice, but domestic heavyweights rapidly stepped in. Alibaba's Qwen picked up major enterprise clients throughout 2024, while DeepSeek stunned the industry by matching Western frontier performance at a fraction of the cost, kicking off a race among local automakers and appliance brands to bake domestic AI into their hardware.

Why it matters

Three years in, ChatGPT has achieved infrastructure-level status faster than almost any consumer software in history. 800 million weekly users prove that generative AI is no longer a parlor trick; it is how millions of people work, learn, and cope with administrative bloat every single day.

However, OpenAI's global playbook shows clear vulnerabilities. We believe the linguistic blind spots in major foundation models are actively driving model fragmentation. When a model hallucinates in Thai or Swahili, it forces entire regions to build sovereign AI to protect their cultural data and local utility.

At the same time, the competitive landscape is shifting from pure benchmark dominance to cost and accessibility. DeepSeek's low-cost breakthrough and the proliferation of nimble, local LLMs show that the Global South will not just passively consume Silicon Valley APIs. OpenAI kickstarted the revolution, but the next three years will be defined by who can make AI cheap, accurate, and culturally native to the other seven billion people on Earth.