OpenAI нанимает «директора по выживанию»: Сэм Олтмен готовится к пришествию GPT-5

Sam Altman is making another move that makes the industry nervously adjust its ties. The appointment of Dylan Scandiano to the position of "Chief Preparedness Officer" looks like OpenAI's attempt to prove they still control the situation. While we entertain ourselves generating pictures and ask a chatbot to write code for a calculator, inside the company they are preparing for scenarios that once seemed like the plot of a Hollywood blockbuster.

Altman speaks openly: the coming models will possess extraordinary power, and old security protocols simply won't handle the load. This appointment does not happen in a vacuum. After high-profile departures of key figures from the safety team, including Ilya Sutskever and Jan Leike, OpenAI's reputation on ethics has noticeably shaken.

Critics accused the company of trading caution for profit and development speed. Scandiano must now become that very lightning rod that will calm regulators and the public. His task is not simply to check code for errors, but to build a system of protection against serious risks that Altman prefers to discuss evasively, but anxiously.

The job title itself is interesting. "Preparedness" is not about passive observation, but about active crisis preparation. In the context of AI, this could mean anything: from preventing the creation of biological weapons with the help of a model to controlling attempts by the system to break out of the digital sandbox.

We are entering an era when the capabilities of neural networks begin to outpace our understanding of how to manage them. OpenAI is effectively admitting that their future products could be dangerous by nature if left unsupervised by specially trained people. For the industry, this is a clear signal: the arms race is moving to a new level.

If before companies competed in the number of parameters and quality of responses, now an important asset becomes the architecture of control. Investors and partners want to be sure that the next breakthrough will not be the last for humanity, or more prosaically, will not lead to billion-dollar lawsuits. Altman understands that without public trust, it will not be possible to scale technology to the level of global infrastructure.

However, behind beautiful words about the benefit to society lies a pragmatic calculation. By creating such positions, OpenAI secures for itself the status of a leader who sets the rules of the game themselves. If Scandiano manages to build a working system, OpenAI will become a standard not only in technology but also in regulation.

This will allow the company to dictate terms to governments, explaining exactly how AI should be controlled so as not to, allegedly, harm progress. A smart move that combines concern for the future with classic market capture. The main point: OpenAI is preparing the ground for launching models whose capabilities could frighten an unprepared user.

Will the appointment of one person be enough to contain the digital genie, or is this just a clever PR move to calm critics before the GPT-5 release?