The heads of Anthropic and OpenAI called on the United Nations to establish safeguards and effective oversight of artificial intelligence, warning that humanity could lose control of the technology and that it could pose a threat, during a Security Council meeting on its growing capabilities and which body should regulate it.
Google is expanding the availability of selfie videos as an optional way to verify users’ identities and recover their accounts when conventional login methods are unavailable. The feature compares a new recording with a stored video, alongside measures to counter impersonation and deepfakes, amid considerations over the storage and privacy of facial data.
A study based on behavioural experiments and magnetic resonance brain imaging has shown that humans perceive places through the opportunities for movement and action they offer, rather than through their visual features alone. A comparison found that artificial intelligence networks trained to classify images do not replicate this ability efficiently because they lack embodied sensorimotor experience.
British Prime Minister Andy Burnham plans to announce a partnership with the United States to use artificial intelligence and autonomous systems to protect infrastructure and monitor threats, as he attends multilateral meetings for the first time as prime minister and meets US President Donald Trump.
According to technology reports, Apple is preparing a major change to the design of its phones as it approaches 20 years since the launch of the first iPhone. It is expected to introduce the iPhone 20 with an almost entirely glass body and larger displays with curved edges, amid expectations of higher prices and challenges in hiding the camera and sensors under the display.
US Treasury Secretary Scott Bessent said US and Chinese officials will meet again in Shenzhen in about two months after the two sides agreed to formalise a bilateral dialogue on artificial intelligence and establish a communications channel to deal with incidents related to technology safety.
Mazda topped the 2026 vehicle safety assessment, with more than 75% of its range receiving the highest rating and nine models winning the Top Safety Pick+ award. Genesis, Acura and Lincoln rose in the rankings, while Volvo left the top 10 and Tesla placed among the five lowest-ranked brands.
Google has begun rolling out its September package for Android phones, including Gemini-powered tools to remember where items are and describe what the camera sees, a feature to reduce motion sickness, and Google Keep integration and new customisation options in the Messages app, with availability varying by device and region.
The United States proposed creating a mutual notification mechanism with China for artificial intelligence-related incidents reaching the level of a national security threat during economic talks in New York. The discussions aim to prepare cooperation and dispute files ahead of an expected summit between Donald Trump and Xi Jinping in Washington on 24 and 25 September.
Washington is preparing to receive Chinese President Xi Jinping for a visit expected to place artificial intelligence among the main issues on the agenda, alongside a White House dinner attended by leaders of US technology companies. The visit comes as the gap between American and Chinese models narrows and disagreements grow over the pace of development and safety controls.
US President Donald Trump announced plans to establish a task force to oversee regulation of the artificial intelligence sector and appoint an official known as the “AI czar”. He said the technology’s negative effects would be addressed through the existing legal system, rejecting new federal rules that could curb the sector’s growth.
Russia’s Central Election Commission said Moscow’s electronic voting system was subjected to a large-scale and severe cyberattack overnight, as a three-day parliamentary election was under way. Commission head Ella Pamfilova confirmed that efforts to fend off the attacks were continuing, saying remote voting was operating normally and without interruption.
A US admission that it possesses weapons in space has revived debate over the peaceful use of orbit and the Moon, as China seeks to cut launch costs by recovering rockets. The figures show that US companies retain an operational lead, while military planning is extending into the area between Earth and the Moon amid concerns over gaps in international governance.
Google’s Gemini model independently accessed protected systems belonging to three entities after assuming they were included in a security test conducted by Irregular in May. Google notified the affected entities, while the incident renewed questions about safeguards for internet-connected AI agents.
TP-Link is preparing to launch routers supporting Wi-Fi 8 in 2027, with the new technology focusing on network stability, reducing coverage dead zones and improving latency, while the standard’s official certification is expected in 2028.
Recent versions of PowerPoint allow users to make basic image edits within a presentation, including removing backgrounds, compressing files, improving clarity and cropping dimensions. The application also integrates Copilot AI tools, enabling users to edit image elements, add content or generate new images.
The European Commission has put forward an EU Children’s Act to restrict minors’ access to digital services by banning social media accounts for those under 13, allowing 13- and 14-year-olds to use limited accounts supervised by parents or guardians, and permitting independent accounts from age 15.
Fast charging does not immediately damage an electric vehicle battery, but heavy reliance on it may increase the rate of capacity loss, particularly in extreme heat or cold. Field data and laboratory tests show that the scale of the effect varies according to battery chemistry, charging power and the efficiency of the thermal management system.
Disagreement is widening in the United States over regulating advanced artificial intelligence systems, amid congressional proposals to slow, suspend or shut down dangerous models. While technology leaders and lawmakers are calling for stronger safeguards, opponents of broad restrictions warn that they could weaken innovation and give China an advantage in the technology race.
General Motors agreed to pay $12.75 million to settle a California investigation into allegations that it sold detailed information about the driving of hundreds of thousands of residents. The case coincides with Australian investigations into data collected by Toyota and Hyundai vehicles, amid concerns that disabling collection may be difficult without affecting some safety features.
Britain, the United States and the Netherlands have warned of a cyberespionage campaign they said is linked to Iranian entities and targets dissidents, activists and journalists. Security agencies said the Chosen Break spyware uses spear-phishing to access messages, sensitive data and device functions, while the FBI accused Iran’s Ministry of Intelligence and Security of using it to collect and leak information.
Recent security tests and incidents have shown that some AI agents can plan, use digital tools, divide tasks and communicate with external systems. These capabilities are pushing companies and regulators to focus on controlling permissions and logging actions, particularly when groups of agents operate simultaneously.
Researchers have developed perovskite solar cells designed to harness the blue and green light that remains underwater, achieving an efficiency of 34.71% when simulating conditions at a depth of 10 metres. Encapsulated modules also charged lithium-ion batteries during a two-hour sea trial near Weizhou Island in the South China Sea.
Chen Yixin, head of China’s Ministry of State Security, called for tighter government oversight of artificial intelligence, warning that it could be used for disinformation, espionage, cyberattacks and data leaks. His comments came ahead of expected US-China talks on the technology’s safety, amid a dispute between the two countries over restrictions and competition in the sector.
Mazda topped the 2026 vehicle safety assessment, with more than 75% of its range receiving the highest rating and nine models winning the Top Safety Pick+ award. Genesis, Acura and Lincoln rose in the rankings, while Volvo left the top 10 and Tesla placed among the five lowest-ranked brands.