A new artificial-intelligence model is so powerful that the White House has asked for its release to be restricted. It's GPT-5.6 by OpenAI, whose advanced cyber capabilities have raised concern in both Washington and on Wall Street.

According to reports, authorities fear the model could create „unprecedented security risks." So OpenAI accepted a temporary solution: instead of a public launch, the model will be available only to government-approved partners while a regulatory framework is developed. In other words - the most powerful technology goes to the state first, not the public.

This isn't an isolated case. The same move was already applied to Anthropic for its models. When two leading AI companies, in the same period, receive a state order to hold back their most advanced models, it tells you Washington already views artificial intelligence as a weapon, not a product.

OpenAI's CEO, Sam Altman, tried to soften the picture: „We've told the US government clearly that this is not our desired long-term model and we'll work with it and others in the industry on a more sustainable approach in the future." Translation: the company would love to sell freely, but for now it has to listen.

For the rest of the world, the Balkans included, the lesson is sobering. The most powerful tools of the future are born behind closed doors, and access to them is decided by a few governments and a few companies. By the time the technology reaches us, it will already have been shaped by someone's security interests - and we, as usual, will get the version someone else decided we're allowed to see.