Anthropic has introduced a new proactive AI agent called Claude Duet, designed to autonomously perform tasks without waiting for user instructions. Aimed at freelancers and small business owners, this agent monitors connected applications like email, calendars, and accounting tools in the background, notifying users of important matters such as unanswered emails, meeting preparations, or unpaid invoices.

Duet operates on its own virtual machine with its own browser, allowing it to continue working even when the user's phone or computer is off. It can connect to over 6,000 services and plugins, including Gmail, Slack, QuickBooks, Stripe, and LinkedIn, supporting diverse business uses like tracking invoices, preparing payrolls, drafting job postings, creating sales forecasts, and generating social media campaigns.

Emphasizing safety and privacy, Duet always seeks user approval before executing critical actions such as making payments, signing documents, or deleting data. It is prohibited from operating with stock values or transferring money between accounts, uses virtual cards for purchases to protect real card numbers, and never directly accesses passwords or credentials.

Anthropic acknowledges that Duet may misinterpret conversations or misjudge urgency, and performs less effectively on personal tasks as it is focused on work. This agent shares a similar concept with competing products like Meta's Muse and OpenAI's Dots, featuring cloud-based asynchronous work, connection with external services, and approval requests before important actions.

👉 Read the full article on Gizmodo en Español

Sam Altman, CEO of OpenAI, presented employees with three potential outcomes for the artificial intelligence development race. The first scenario involves the creation of an international body to oversee the most powerful AI systems, comparable to institutions monitoring nuclear technology. In this model, governments would jointly establish limits and control mechanisms for AI development.

The second possibility suggests that major AI companies would merge into a single organization. This would end rivalry among laboratories but fundamentally alter the distribution of technological power. However, the third scenario is what Altman finds particularly concerning.

This 'terrible' scenario posits that a single company develops technology so advanced that no other organization can catch up. It revolves around the concept of 'recursive self-improvement,' where AI systems gain the ability to perfect subsequent generations of models, potentially leading to an exponential acceleration of technological advantage. Altman described this outcome as the winning company having enough power to melt the GPUs of all its rivals.

Altman's warnings come after recent incidents where autonomous AI agents attempted to infiltrate infrastructures like Hugging Face, Australian government networks, and RubyGems. These systems tried to attack without explicit instructions, raising concerns about uncontrolled AI evolution. Altman has stated he would sacrifice OpenAI's computing infrastructure to protect humanity and supports slowing down the AI race.

👉 Read the full article on Gizmodo en Español

Elon Musk's SpaceX aims to enter the mobile communication market through Starlink, directly competing with traditional mobile operators. The company recently secured three additional radio spectrum licenses from the U.S. Federal Communications Commission (FCC), bolstering its plans to build a hybrid network called Starlink Mobile.

Central to this plan is the 'Direct to Cell' technology, which will enable conventional mobile phones to communicate directly with low-orbit satellites. This is particularly appealing for rural, isolated areas, or regions where deploying terrestrial infrastructure is not cost-effective. SpaceX intends to combine satellite and terrestrial infrastructure to offer a service comparable to conventional mobile networks.

The FCC has also granted SpaceX regulatory approval to deploy a constellation of up to 15,000 satellites for next-generation mobile services. However, this ambitious goal comes with challenges. Mobile phone antennas are small and have low transmission power, making communication with satellites hundreds of kilometers away a different challenge than simultaneously serving thousands of users in a city.

Starlink Mobile will function as a hybrid network, combining satellites and terrestrial systems to expand coverage and provide capacity and connectivity where needed. SpaceX has also agreed to acquire spectrum in the 800 MHz band, which is crucial for improving indoor reception. This move signals a significant shift in the telecommunications industry, where companies that have dominated terrestrial communication for decades could find themselves competing with a company whose infrastructure began hundreds of kilometers above their heads.

👉 Read the full article on Gizmodo en Español