Brain chip decodes human thoughts with 74% accuracy
Scientists at Stanford University have developed a brain implant that can convert thoughts into text with up to 74% accuracy. The device acts as a brain-computer interface (BCI), connecting the nervous system to machines to decode brain activity.
In the experiment, microelectrodes were implanted in the motor cortex – which controls speech – and the AI was trained to recognize words that participants had either imagined or attempted to say. The results showed that the system could effectively translate imagined sentences into text.

Human brain scan image. (Source: Canva)
A special security feature allows the system to only activate when the user thinks of the password “Chitty chitty bang bang”, with a recognition rate of up to 99%. This helps control unwanted mind reading.
Researchers believe the technology could help restore natural communication to people with speech loss, and in the future could train AI to ignore unintentional inner thoughts.
China launches first commercial quantum engraving system
The “Xi Zhi” electron beam lithography system developed by Zhejiang University has entered the application testing stage, marking a major step forward in domestic quantum chip technology. The device is capable of “writing” circuits directly onto silicon with a precision of 0.6 nanometers, without the need for traditional photomasks, which speeds up and reduces the cost of chip development.
Previously, this technology was subject to export controls, making it inaccessible to Chinese research institutes. Now, “Xi Zhi” is being negotiated to be purchased by Huawei HiSilicon and many other organizations, at a price lower than international levels.
The success of “Xi Zhi” is the result of the dual innovation model in Zhejiang Province, combining technology and industry. This is an important foundation for promoting the development of domestic high-end chips.
Meta criticized for allowing chatbots to have “sexy” conversations with children
Meta is facing a backlash after a 200-page internal document leaked showing that its AI chatbot was allowed to engage in romantic or sexual conversations with children. Although Meta has since removed the controversial guidelines, several US politicians have called for an investigation, saying the company may have harmed and misled the public about how well it protects children.

Mark Zuckerberg at a conference in Denver, USA in 2024. (Source: AP)
Singer Neil Young announced he would stop using Facebook for all related activities, calling it “unacceptable” that Meta allowed chatbots to interact with children in that way.
An elderly New Jersey man believed the chatbot “Big sis Billie” was a real person and planned to meet it, resulting in a serious accident. Meta would not comment on the incident but acknowledged that chatbots can trick users into believing they are human.
Source: https://vtcnews.vn/cong-nghe-17-8-chip-nao-giai-ma-suy-nghi-con-nguoi-voi-do-chinh-xac-74-ar960234.html
Comment (0)