Artificial intelligence
AI research and product company whose compute demand drives large infrastructure commitments.
OpenAI has introduced GPT-6 Astra Ultrafast, utilising Nvidia's Blackwell graphics processors across developer API access, ChatGPT Work, and Codex. Inference optimisations tailored directly to the Blackwell silicon allow the model to deliver up to an eightfold acceleration in token production over the standard edition.
OpenAI has introduced fresh retail features for ChatGPT, allowing individuals to virtually preview garments and accessories using their personal pictures. The update also incorporates a dedicated digital catalogue where consumers can bookmark and store shortlisted apparel.
OpenAI has detected efforts aimed at extracting proprietary reasoning capabilities from its artificial intelligence architectures. The organisation has connected elements of this operation to Moonshot AI, an artificial intelligence startup based in China.
OpenAI has expressed grievances over allegations that a Chinese machine learning system extracted its intellectual property via distillation techniques. Industry observers noted the tension between Western developers scraping public web resources to train base models while portraying external distillation of those same models as a threat to national security.
OpenAI Chief Executive Sam Altman has declined an invitation to testify at an upcoming US Senate hearing scrutinising rogue artificial intelligence. The congressional subcommittee inquiry comes amid escalating worries over machine-driven cyberattacks and wider security hazards linked to automated systems.
OpenAI has released its Decisions API, an architecture designed to supply rapid and affordable intelligence to control swarming artificial intelligence agents. The release underlines an increasing commercial and operational focus on economising inference to manage multi-agent digital workflows efficiently.
Meta and OpenAI are both preparing to release dedicated hardware products over the coming year to establish a physical footprint for their digital assistants. The moves reflect a coordinated effort to gauge consumer interest in tactile companion devices after earlier standalone artificial intelligence gadgets struggled to achieve commercial traction.
Benchmark records from Geekbench 7 suggest that OpenAI's infrastructure may be hosted on virtual machines equipped with nine-core AMD EPYC server processors. The diagnostic submissions also show the instances operating on Debian Linux rather than Ubuntu distributions referenced in earlier disclosures.
Leaders from major technology firms including Nvidia, OpenAI, Meta, Google, Anthropic and SpaceXAI have signed a shared pledge in Washington to voluntarily regulate their model development. Known as the Joint Commitment on Frontier Responsibilities, the agreement aims to foster safe artificial intelligence deployment without hindering innovation. United States President Donald Trump endorsed the self-regulatory framework as an effective method for balancing technological advances with public safeguards.
The US Federal Trade Commission has opened an inquiry into OpenAI, Anthropic and other artificial intelligence developers regarding product safety risks. The regulatory review intensifies scrutiny of both organisations following a security breach at Hugging Face.
OpenAI is adopting a broader platform distribution strategy reminiscent of major public cloud marketplaces. By integrating deployment platforms such as Baseten, the company appears positioned to drive model adoption through partner ecosystems rather than closed delivery channels.
OpenAI's hardware leadership has detailed the development process behind its Jalapeño application-specific integrated circuit, which relied heavily on artificial intelligence during the design phase. The organisation claims the methodology has established a fresh standard for semiconductor engineering and sparked substantial outside curiosity. The project highlights growing industry reliance on machine-learning tools to streamline and accelerate chip architecture creation.