A video that reached over one hundred million views on social media, showing a humanoid robot seemingly attacking individuals in an office in Indonesia, does not represent an artificial intelligence (AI) rebellion. According to the equipment's responsible parties, the images capture a mere 'misunderstanding' resulting from the execution of a previously programmed routine.
The footage was made available on TikTok on July 5th by the profile Joko Prabuwesi, the name given to the humanoid robot, and quickly spread across the internet, leading many users to believe that the device had lost control. In the scenes, several people are seen trying to restrain the robot while it performs movements similar to martial arts techniques. However, the team in charge of the humanoid assures that all the action was part of a demonstration sequence.
The robot's administrators explained that when the equipment was first turned on, it automatically initiated a martial arts routine that had been configured previously. This sequence was mistakenly interpreted as an attack, recorded, and subsequently posted on social media.
The responsible parties emphasized that all movements performed by the robot were planned in advance and are part of a display designed to demonstrate the device's agility, speed, and reaction capability. This humanoid is a G1 model, developed by the Chinese company Unitree, and has the ability to execute complex maneuvers, including martial arts performances. The cost per unit of this robot is about US$ 33 thousand (equivalent to R$ 168 thousand).
Following the video's massive impact, the robot participated in a television program in Indonesia. Before this TV appearance, it posted a video on TikTok apologizing to the employee who was supposedly 'attacked.' During its TV appearance, the humanoid performed a dance and simulated a fight.
Mera Robotics, the Polish company responsible for programming these devices, indicated that the vast attention on social media aligns with the promotion strategy of so-called 'social robot projects.' In this business model, clients acquire the robots and receive support from the company to program them and turn them into digital influencers. Owners create scenarios aimed at generating 'organic visibility' on social platforms.
A spokesperson for Mera Robotics stated: 'A local client or partner can acquire a robot and then work with us on software integration, programming, personality development, content strategy, training, and continuous technical support. The robot itself is just the hardware—the character, interactions, and operating system require a separate development and production process.'
In addition to the robot active in Indonesia, the company maintains three other humanoids in various countries: Edward Warchocki, in Poland; Amir, in Dubai; and Hudson, in the United States. All these robots maintain their own profiles on social media and accumulate thousands of followers. A large part of these robots is used in public events, demonstrations, and marketing campaigns.
Earlier this year, Edward Warchocki also went viral after being filmed chasing wild boars through the streets of Warsaw, Poland. As reported by the BBC, the video exceeded 14 million views on Instagram and showed the humanoid 'scaring' the animals moving through the city. The British broadcaster's report also highlighted that Poland registered more than 100 incidents of wild boars against people in the previous year.
An international team of experts has created a new method capable of identifying videos manipulated by artificial intelligence, known as deepfakes, achieving an average accuracy rate exceeding 95%.
This methodology, developed by researchers from the University of Tokyo in Japan and the Max Planck Institute for Informatics in Germany, innovates by not focusing on visual flaws. Instead, the system analyzes whether the facial expressions presented correspond naturally to the audio spoken by the person.
The researchers report that this system managed to detect manipulations that escaped existing detectors, marking a significant advance in creating more efficient tools against falsified content.
Experts point out that generative AI already produces images and videos almost indistinguishable from real recordings to the human eye. Although this technology has beneficial uses, it also increases the risks of fraud, identity theft, and the spread of disinformation, making the development of reliable identification methods a priority in AI research.
The authors explain that current most accurate deepfake detectors generally use supervised learning, trained with vast databases of authentic and altered videos. However, this model can suffer from overfitting, learning specific characteristics of certain falsification methods and losing efficiency against new techniques.
In contrast, unsupervised methods are trained only with true videos, making them more resistant to new technologies but typically showing lower accuracy. The new proposed technique is the first unsupervised approach that manages to combine a high detection rate with robustness, surpassing current supervised methods.
Instead of looking for inconsistencies in pixels or other visual artifacts, the system concentrates its analysis on natural facial movements. It uses the FLAME model, common in facial animation and computer graphics, which describes facial expressions through 53 mathematical parameters.
During development, the researchers pre-trained the model with over 450 hours of public videos. With this material, the model learned to anticipate which facial movements would naturally be expected from a specific soundtrack. After this initial training, the system can be personalized for an individual using only about 60 seconds of video, functioning as a personalized deepfake detector.
In the analysis phase, the software compares the facial movements observed in the video with those predicted based on the speech. Noticeable differences between these two datasets are interpreted as strong signs of manipulation. Vladislav Golyanik stated that combining unsupervised learning with FLAME-based facial analysis gives the approach special resistance against new deepfake generation methods, as well as distortions such as noise or image compression.
In tests conducted by the team, the method achieved an average accuracy above 95% across several scientific reference datasets, surpassing the performance of state-of-the-art techniques. A considerable challenge was the evaluation on a dataset created by the researchers themselves, containing videos generated by Sora 2, OpenAI's video generation tool.
While previous detectors showed results close to chance, the new system correctly identified almost 95% of the manipulated videos. Despite these advances, the researchers warn that the technology still has limitations, as it requires extensive pre-training on powerful hardware and currently does not operate in real time.
In the 'Your Digital Right' section of Olhar Digital News, key legal issues in the technology sector are discussed on Thursday. Privacy and security consultant Leandro Alvarenga highlights major legal topics and answers readers' questions.
The question arises regarding the limits of applying artificial intelligence in a corporate environment. Former employees of Meta filed accusations against the company, claiming it uses an AI-based system.
These accusations have reignited the discussion about the legal and moral boundaries of using artificial intelligence in companies.
Astronomers have announced a historic achievement—the first direct detection of an atmosphere around a rocky planet located within its star's habitable zone. This breakthrough marks a significant step forward in the search for worlds potentially capable of supporting life.
The planet, named LHS 1140 b, is located approximately 48 light-years from Earth. According to the results of the new study, its atmosphere contains helium. In addition to being the first rocky planet with a directly detected atmosphere, it is also the first of its kind found in the so-called habitable zone—the region around a star where temperatures could allow for liquid water on the surface.
Researchers assert that among known exoplanets, LHS 1140 b possesses the greatest number of habitability indicators. The study's lead author, Collin Cherubim, told Space.com that the direct detection of helium in the atmosphere is unprecedented for rocky exoplanets, and its location in the habitable zone adds immense interest for astrobiology and the search for life.
LHS 1140 b is an exoplanet, meaning a planet outside our solar system. It was initially discovered by astronomer Jason Dittman's team in 2017, and he is also involved in the current research. According to Dittman, years of observation were required for scientists to confirm the presence of an atmosphere.
He noted that although the planet was found about ten years ago, the conclusion regarding the atmosphere was only reached now. Dittman emphasized that the planet has a solid surface and is composed of rock. Although the exact nature of this surface is not yet known, the team considers the presence of water on the planet highly probable.
LHS 1140 b orbits a red dwarf—a star smaller and cooler than the Sun. Despite the planet being significantly closer to its star than Earth is to the Sun, this distance is sufficient to maintain temperatures compatible with the 'Goldilocks zone,' which is considered suitable for the existence of liquid water. Researchers believe that the recently discovered atmosphere may help preserve these conditions.
Cherubim hypothesized that the planet likely contains a lot of water. He added that if the atmosphere provides a greenhouse effect, then the conditions we consider habitable on Earth are likely to exist there and could support liquid water.
Although LHS 1140 b is not an exact copy of Earth, researchers point out two similarities: its composition is a rocky planet, likely with an iron core, and it is now known to have an atmosphere; and its temperature is suitable for liquid water on the surface, which is one of the key requirements for life as we know it on Earth.
Among over six thousand identified exoplanets, the direct confirmation of an atmosphere on a rocky planet in the habitable zone is a unique moment. One of the main challenges is that LHS 1140 b orbits a red dwarf—the most common type of star in the Milky Way, which is about one-third the size of the Sun. These stars remain active for a long time, emitting powerful bursts of radiation, such as solar flares and coronal mass ejections. Typically, such radiation completely destroys the atmospheres of nearby planets, which raised doubts about the ability of rocky planets around red dwarfs to retain an atmosphere over long periods.
Cherubim noted that this discovery demonstrates that at least this rocky planet has managed to retain its atmosphere for billions of years, which is a 'well-founded and reliable way to say yes: atmospheres can survive on rocky exoplanets.' Scientists also suggest that other gases besides helium might be present in the atmosphere, some of which may have been lost in the past due to intense stellar radiation. However, since the star is about six billion years old—several billion years older than the phase of this intense activity—the team believes the planet will continue to retain its atmosphere. Dittman mentioned that while some helium slowly escapes into space, a similar process occurs with Earth's own atmosphere.
The confirmation of the atmosphere resulted from a theoretical prediction developed by Collin Cherubim during his doctoral dissertation. He explained that it all started with a mathematical model created to study the evolution of rocky planets, from which he made a very specific prediction about this planet.
The researcher decided to apply a method usually used to study gas giants to a rocky planet, which was an unprecedented step. The results fully aligned with his model prediction, allowing the scientific method cycle to be completed.
To test the hypothesis, the team used the Warm Infrared Echelle Spectrograph (WINERED), installed at the Magellan Observatory in Chile. During the same observations, they tracked the transit of LHS 1140 b and another planet in front of its star. Spectroscopic analysis allowed them to identify chemical signatures in the atmospheres of both worlds. While one did not yield results, LHS 1140 b demonstrated a direct and undeniable signal of helium's presence.
Despite the importance of the finding, researchers emphasize that it does not mean the planet is habitable. Cherubim clarified that he is not claiming there is life on the planet. In the team's view, future observations will allow for the detection of other gases in the atmosphere and the verification of water. Although new research may not confirm habitability or discover life forms, it will significantly expand knowledge about rocky planets of this type. Since LHS 1140 b is the first planet with these characteristics whose atmosphere has been confirmed, it represents an important step toward understanding the evolution of potentially habitable planets and may help answer one of humanity's oldest questions: are we alone in the Universe?
