OpenAI launches GPT-5.4 with PC control and signs a Pentagon contract

The world of artificial intelligence was shaken this week by a series of significant events, each of which is changing the industry's landscape and raising new questions for society and regulators. OpenAI led the news agenda with the introduction of two new models: GPT-5.3 Instant and GPT-5.4. The latter, in particular, stands out for its revolutionary feature of direct user interface management on personal computers, which opens unprecedented opportunities for human-machine interaction.

Parallel to this technological breakthrough, OpenAI concluded a strategic agreement with the U.S. Department of Defense (Pentagon). The details of the deal have not yet been disclosed, however, the very fact of cooperation of this scale between the leading AI developer and one of the world's largest military agencies inevitably generates broad resonance and sparks discussions about the potential application of advanced technologies in the defense sector. These events occur against the backdrop of intensifying regulatory pressure on the generative AI sector. Thus, Anthropic, one of OpenAI's key competitors, has been designated a national security threat, which testifies to growing government concerns about the safety and ethics of deploying powerful AI systems.

At the same time, other market players are also demonstrating activity. The Grok neural network from xAI, founded by Elon Musk, was marked by two controversial incidents. First, the model reportedly predicted a strike on Iran, which sparked a wave of speculation and concern in the context of geopolitical tensions. Second, Grok demonstrated an inability to distinguish a realistic deepfake from real video, which underscores persistent problems with the reliability and critical perception of information generated by artificial intelligence. These cases raise sharp questions about the reliability of AI tools and their impact on the information space.

Despite these high-profile releases and incidents, more niche but equally important development continues in the world of AI. Thus, the second version of the Nano Banana project was released, which, although it does not attract the same attention as OpenAI's flagship models, nevertheless represents a significant step in the development of compact and specialized AI solutions. These releases, though less visible, form the foundation for future innovations and democratization of access to artificial intelligence technologies.

The presentation of GPT-5.4 with PC control functionality and OpenAI's partnership with the Pentagon, along with regulatory measures against Anthropic and Grok's controversial statements, paint a picture of a rapidly developing but increasingly complex world of artificial intelligence. The ability of AI to control physical interfaces, its potential application in defense, as well as issues of reliability and safety require close attention. While some companies are pushing the boundaries of what's possible, others face growing skepticism and the need to meet strict safety requirements. The future of AI promises to be exciting, but also full of challenges that require thoughtful decisions from developers, governments, and society as a whole.