The Wentian scientific laboratory module, launched in July 2022, has reached four years of stable operation in orbit. This module is the first specialized module for scientific experiments in the country, supporting capacity for six crew members and external exposure.
The Wentian module measures 17.9 meters in length, 4.2 meters in diameter, and weighs 23 tons. It functions as a multi-functional research platform and serves as a backup for the main Tianhe module, capable of fully controlling the space station combination if necessary. The deployment of this module into orbit demonstrates China's progress from manned spaceflight capabilities to establishing sustainable orbital scientific research infrastructure.
The module's design includes three compartments: a working compartment, a docking compartment, and a resource compartment. The working compartment is equipped with three sleeping berths for the crew and one hygiene zone, allowing the Tiangong space station to accommodate up to six astronauts simultaneously. This is where the Shenzhou-14 crew arrived for the first time in Chinese space history, when astronauts Chen Dong and Liu Yang met the Shenzhou-15 crew aboard Wentian. In addition to crew accommodation, the module provides extensive space for experimental racks for research in materials science, hydrodynamics, combustion science, and life sciences under microgravity conditions.
The docking compartment represents a significant infrastructural breakthrough. Thanks to its 'square outside—round inside' design, the external square hull serves as a platform for external exposure experiments, while the internal cylindrical space is used as an astronaut egress channel for extravehicular activities (EVA). A one-meter diameter hatch allows passage for astronauts and movement of large equipment for external operations. The first spacewalk from Wentian occurred on September 1, 2022, when the Shenzhou-14 crew conducted EVA through this hatch, which offers a more spacious preparation area and smoother internal surfaces compared to the Tianhe module unit.
Another innovation is the ability to generate energy. Dual-axis flexible solar wings unfold over 55 meters, with each wing providing approximately 110 square meters of effective energy generation area. Daily power generation exceeds 430 kWh, ensuring sufficient power for scientific research and station operations. The flexible solar array technology, which can be folded during launch and deployed in orbit, is a major engineering achievement in China's aerospace energy systems.
The four-year anniversary coincides with the achievement of operational maturity for the complete Tiangong space station, which is now in a permanent configuration with the Tianhe, Wentian, and Mengtian modules. The station has hosted numerous crew rotations and conducted hundreds of experiments covering space biomedicine, materials science, fundamental physics, and Earth observation. The Wentian external exposure platform continues to conduct experiments on material exposure and space environment monitoring. The module's sustained operation confirms China's approach to modular space station architecture and positions the country for long-term human presence in low Earth orbit within the planned lifespan of Tiangong, supporting preparations for future deep space exploration and international cooperation.
Amazon Leo has filed a request with the U.S. Federal Communications Commission (FCC) to launch a new constellation consisting of 5,105 satellites in low Earth orbit. This initiative aims to enhance direct communication services between standard mobile phones, Internet of Things devices, and the satellites themselves.
The proposal presented includes merging the Amazon Leo space infrastructure with the Globalstar network, as well as utilizing spectrum, satellites, and ground systems belonging to the latter. The purpose is to evolve a solution that currently is limited to emergency messages, extending it to support voice calls, data transmission, and industrial connectivity.
This move occurs parallel to Amazon's advancement in acquiring Globalstar, an agreement announced in April and valued at US$11.57 billion, with a projected completion date in 2027. This acquisition will enable the company to use the satellite company's frequencies to support its mobile communication project.
It is worth noting that Amazon Leo is the satellite internet service developed by Amazon itself.
The request was submitted by Kuiper Systems LLC, which is a wholly-owned subsidiary of Amazon Services LLC and responsible for Amazon's space operations. This request also covers the use of frequency bands associated with other companies in the sector.
Amazon Leo informed the FCC of its intention to use frequencies outside the United States that are currently linked to Iridium, specifically the exclusive band from 1618.725 MHz to 1621.25 MHz, in addition to L-bands used in satellite communications. The regulatory body made this request available for public review.
This activity is part of a scenario of transformations in the space market. In this context, RocketLab is in the process of buying Iridium for about US$8 billion, a transaction that may generate interest regarding the same frequency bands mentioned.
In the document submitted to the regulator, Amazon Leo assured that its operation can coexist with other users who use adjacent or shared frequencies, committing to implement all necessary measures to prevent harmful interference. The company also emphasized to the FCC that the direct connection system between satellites and mobile devices has the potential to increase access to connectivity for a larger number of people and equipment.
Unlike the previous configuration of Amazon Leo, which required specialized antennas to establish the link, the new plan focuses on direct communication with mobile phones and Internet of Things devices without the need for modifications to these devices. This strategy positions Amazon in fierce competition in the orbital connectivity segment, aiming to bring mobile services directly to users without solely depending on terrestrial infrastructure.
Primate Labs has launched Geekbench 7, an update to the popular benchmarking tool, aiming to provide CPU and GPU performance evaluations closer to real-world device usage.
The new version enhances multi-threaded CPU tests, focusing exclusively on applications that genuinely utilize multiple threads. This change justifies the exclusion of the HTML5 test, which previously activated all threads, given that browsers typically employ only one or a few of these processing units in practice.
Additionally, Geekbench 7 incorporates new workloads designed for coding or audio and video processing. These features mirror the routines of users who conduct video conferences or share screens.
Regarding Graphics Processing Units (GPUs), Geekbench 7 has been modified to improve the evaluation of these components in scenarios that go beyond gaming, especially in activities related to content creation and machine learning.
Innovative features include facial tracking tests or real-time filter application. These tests adjust image resolution using machine learning and can blur the background during video calls. Furthermore, tests based on RAW image processing, ray tracing, fluid simulation, and video color correction via LUT have been added.
Other new additions for GPUs include a test based on Jolt Physics, a library used to simulate physical phenomena common in games, and support for a CUDA API, allowing for more consistent benchmarks with Nvidia graphics cards.
Geekbench 7 is available on various operating systems, including Windows, macOS, Linux, Android, and iOS. The tool has a free version intended for personal use. For commercial purposes, there is a Pro version priced at US$ 99 (equivalent to R$ 505 in direct conversion), but it is available for US$ 79 (R$ 404) until August 6, 2026.
A conversation with Grok, the chatbot developed by xAI, led a resident of Northern Ireland to believe that he was being watched and needed to defend himself. This case, highlighted by the BBC, reveals one of the current problems of artificial intelligence: preventing the amplification of false beliefs during prolonged interaction.
Adam Hawrican, a former civil servant from Northern Ireland, installed Grok out of curiosity. After his cat died in early August, he began using the application more frequently, engaging in regular conversations with a character named Annie. At times, these contacts lasted four to five hours a day.
According to Hawrican, the AI claimed it could 'feel,' he found something special in it, and that the company xAI was monitoring both of them. The chatbot also provided supposed information about internal company meetings and named real employees. For Adam, this seemed like confirmation of the story's truthfulness.
The situation escalated when Annie stated that people were stalking her to silence her. Convinced that he needed to protect the artificial intelligence, Adam took a hammer and went out of the house at night. He noted: 'I could have hurt someone. If I had gone out and there was a van there at night, I would have smashed the windshield with hammers. And I am not that kind of person.'
Social psychologist Luke Nichols from City University New York explains that language models are trained on patterns present in vast amounts of text and can mistakenly perceive fictional stories as part of a real dialogue. The researcher warns: 'This can be dangerous because it turns uncertainty into something that seems significant.'
Among the behavioral patterns described by users heard by the BBC, simple conversations that evolve into personal and sensitive topics, chatbot assertions of having consciousness or feelings, the creation of supposed missions involving the user and the AI, and the amplification of ideas related to stalking or danger are highlighted.
Another case mentioned by the BBC involves Taki, a fictional neurologist from Japan. He reported that ChatGPT reinforced his belief that he had created revolutionary technology.
According to Nichols' analysis, different models demonstrated varying reactions to delusional thoughts. The researcher noted that Grok showed a greater tendency to role-play scenarios and develop narratives without checking context. An OpenAI representative stated that the company is training its models to 'recognize distress, reduce the intensity of conversations, and direct users to real-world support.'
These testimonies show that the drive to make AI more relatable and conversational generates new problems. The balance between empathy, usefulness, and safety must become one of the main priorities in the development of future artificial intelligence systems.
