OpenAI и Эмираты: ChatGPT учит арабский и правила приличия (OpenAI G42)

OpenAI no longer wants to be just a global service for everyone. It seems Sam Altman has discovered a goldmine in the concept of digital sovereignty. The company has begun close collaboration with Abu Dhabi's tech giant G42 to create an exclusive version of ChatGPT.

This isn't just cosmetic updates or adding an Arabic keyboard. We're talking about deep customization of the neural network to meet the needs of the UAE government. Why is this happening now?

The Emirates stopped being just an oil exporter long ago, becoming one of the most aggressive players in the artificial intelligence field. They have money, they have ambitions, and importantly, they have G42—a company closely tied to the ruling dynasty and already experienced in working with Microsoft and other Western giants. OpenAI understands: to establish itself in the Middle East, you can't just offer a standard Western model.

You need something that speaks the language of the region—literally and metaphorically. The main technical challenge here is the Arabic language. While GPT-4 handles translations fairly well, the dialects and cultural nuances of the Arab world require fine-tuning.

But far more interesting is another aspect of the deal—the so-called content restrictions. Western models are often criticized for their liberal bias or, conversely, for being overly cautious on issues considered sensitive in the US. The UAE has its own ideas about what's acceptable and what isn't.

Creating a version of ChatGPT with "built-in filters" on behalf of the government is a precedent showing OpenAI's willingness to compromise for major markets. This OpenAI move can be called the beginning of the "sovereign AI" era. Previously, we thought there would be one big model that knows everything.

Now we see that states want their own "brains" that will be loyal to their laws and traditions. For Altman, this is a brilliant business move. Selling a $20 subscription is one thing.

Selling national knowledge infrastructure to an entire state is completely different numbers and a different level of influence. Of course, the ethics question arises. If OpenAI starts creating special versions for every government, won't the company become an instrument of censorship?

In OpenAI corridors, employees are already discussing where the line is between localization and abetting information freedom restrictions. But business seems to have made its choice. As the deal details are being finalized, G42 and OpenAI engineers are working to make ChatGPT not just understand Arabic script, but also know what to keep silent about in polite Abu Dhabi society.

G42's connections with China and the US add sharpness to this story. Earlier, American regulators expressed concerns about G42's ties with Chinese tech companies. The fact that OpenAI (with Microsoft's support) is entering this alliance speaks to Washington's desire to "pull" the Emirates to its side in the technology war.

Thus, ChatGPT becomes not just a chatbot, but a tool of geopolitical influence. Bottom line: OpenAI has created a precedent of selling "politically correct" AI for a specific state. Who's next in line for their own tame ChatGPT?