This week, the artificial intelligence sector has continued to generate headlines, focusing on the recent launches by OpenAI and Anthropic in the health field, as well as Anthropic's new agent, which promises to compete with Gemini developments. However, the most significant news has been the start of monetization for ChatGPT: OpenAI has begun testing ads in both the free version and the GO subscription, accelerating its revenue strategy. The company is estimated to reach $20 billion in revenue this year, anticipating the arrival of advertising in more services, especially in free or low-cost models, while premium plans will remain ad-free. OpenAI's growth is notable, with revenue and infrastructure expenses advancing in tandem, although profits are expected to soon surpass costs. This monetization context coincides with internal tensions in startups in the sector, such as Thinking Machines, founded by Mira Murati, where personal and leadership conflicts have led to the departure of several co-founders, reflecting the intense competition for talent in AI. Apple, for its part, is working on a smart pin with AI, the size of an AirTag, equipped with cameras, microphones, and speakers, which could launch in 2027. This positions it against OpenAI and Meta in the race for personal AI devices, while Google and Samsung prepare their own smart glasses. In research, a Chinese study highlights the use of chatbots to improve triage and medical referrals, successfully reducing consultation time and improving coordination between doctors and patients. This advancement underscores the potential of AI to optimize health systems, especially when developed in collaboration with local professionals. Google has introduced TranslateGemma, an open-source translation model capable of translating 55 languages, including Catalan, with quality comparable to major commercial models. Its modular design allows it to run on mobile devices, laptops, or in the cloud, facilitating integration into various products and services. Regarding advertising in ChatGPT, OpenAI has begun pilot tests with advertisers in the United States, showing contextual ads to both free users and low-cost subscribers. The billing model will be based on impressions, similar to social media, and the company plans to soon enable a self-service platform for advertisers. With nearly 900 million weekly users, the advertising potential is enormous, although the company is proceeding cautiously to avoid affecting user experience. The strategy is led by Fidji Simo, former Facebook and Instacart, reinforcing the ambition to turn ChatGPT into a leading commercial platform. In the business realm, Elon Musk has legally claimed up to $134 billion from OpenAI, arguing that the company should return the profits generated from his initial contribution, although OpenAI has challenged the methodology used to calculate that figure. Regarding hardware, OpenAI plans to launch its first physical device in 2026, in collaboration with Jony Ive and the LoveFrom team, betting on a screenless ambient AI experience focused on voice and contextual interaction, moving away from traditional wearable formats. Anthropic, in partnership with Teach For All, has launched a global initiative to train over 100,000 teachers in AI, facilitating access to tools like Claude and promoting the co-creation of educational solutions tailored to local contexts. Initial results include interactive curricula and gamified applications in various countries, aiming to place educators at the center of digital transformation. Finally, X (formerly Twitter) has re-released the code for its recommendation algorithm, claiming it is based on AI models without manual intervention. However, the measure has been met with skepticism, seen more as a gesture of transparency than a real change, in a context of regulatory pressure and controversies over the use of its chatbot Grok. In research, Google, the University of Chicago, and the Santa Fe Institute have proposed that the most advanced reasoning models function like small internal societies, where different perspectives interact and converge, improving the ability to solve complex tasks. Additionally, the Chinese study on health chatbots confirms that co-design with local professionals is key to success in resource-limited environments.