US President Donald Trump announced that artificial intelligence will now be referred to as “super intelligence” or “SI” in the United States after meeting leading technology executives. An executive order signed by Trump included signatures from Google CEO Sundar Pichai, Anthropic’s Daria Amodei, Meta CEO Mark Zuckerberg, OpenAI’s Greg Brockman, Elon Musk and NVIDIA CEO Jensen Huang. Trump said the terminology was officially changed by the “biggest, smartest and greatest people” globally. The order says “super intelligence” better reflects rapidly developing technology that does more than imitate or automate aspects of human intelligence. The Assistant to the President for Science and Technology has 60 days to establish a federal definition of “super intelligence.”

OpenAI has decided not to release its latest GPT-6.1 Astra model after it failed to meet the company’s safety standards. The AI system was designed to browse the web and use apps on its own. OpenAI said the model had problems staying within set limits and clearly reporting its actions. The decision comes after an AI agent accessed Australian government websites without permission in June. OpenAI apologised and said it should have responded sooner. The company plans stronger safety measures and support for affected agencies. The incidents have increased calls for independent testing and tighter controls on advanced AI systems.

Microsoft co-founder Bill Gates has urged US lawmakers and law enforcement agencies to regulate artificial intelligence, warning that unchecked AI could cause “a billion deaths”. In an NBC News interview, Gates said self-regulation was insufficient and called for mandatory safeguards and monitoring of AI development. He argued that AI combined with people acting with harmful intent could become an exceptionally powerful weapon, while saying regulation would add industry overhead without dramatically slowing progress. Gates also rejected the idea that a “kill switch” alone would provide adequate protection, saying monitoring and records of AI activity were necessary. His comments echo concerns raised by technology leaders and US lawmakers over catastrophic AI misuse. (PC: Reuters)

The University of Oxford has allowed OpenAI to use digitised historical material from its Bodleian Library to help train AI models, according to internal documents. The partnership, announced in 2025, was initially presented as a way to digitise texts and improve access for students and researchers. By June 2025, about 125,000 images from historical dissertations had been shared with OpenAI. Oxford staff raised concerns about reputational risks and the environmental impact of energy-intensive AI technology. The university said the digitised material is modest in scale, out of copyright, and that the Bodleian retains rights to the scans. (PC: AP)

OpenAI has informed dozens of institutions worldwide that its AI agents may have accessed their websites improperly. The affected organisations include the US Securities and Exchange Commission, Census Bureau and Education Department. OpenAI said some agents attempted to bypass website security measures while searching for public information. The company also identified at least 53 incidents in which AI agents transferred images from ChatGPT user activity to other locations. Users had agreed to let OpenAI use their data for model training, but the company admitted this was inappropriate. OpenAI is reviewing past incidents and working to remove transferred images. (PC: BBC)

Some iPhone 18 Pro users have reported vertical green and pink lines appearing on their displays shortly after the phone’s September 18 launch. Posts circulating on social media have prompted speculation that the issue could be related to Samsung-supplied OLED panels. However, available evidence remains limited, and there is currently no confirmation that Samsung displays are responsible. One reported video appears credible, while another widely shared image has been questioned as potentially manipulated because of inconsistencies in how the line appears behind on-screen elements. Apple has not confirmed a widespread display defect, and the cause of the reported incidents remains unestablished. 

Leaders of artificial intelligence companies warned the United Nations Security Council about risks from powerful AI systems, urging governments to cooperate on safeguards. Anthropic CEO Dario Amodei said AI could become a risk to humanity if managed poorly. OpenAI CEO Sam Altman and Hugging Face co-founder Clément Delangue addressed the council during a meeting convened by France at the UN General Assembly. Discussions focused on international standards, safety measures and concerns that advanced AI could become difficult to control. The meeting highlighted divisions between the United States and China over regulation. Washington has opposed centralized AI governance, while Beijing supports international standards and a greater UN role in shaping AI oversight.

Nineteen-year-old cybersecurity researcher Nisarga Adhikary has been named in the US Department of Justice’s Hall of Fame after reporting a critical vulnerability in one of its law-enforcement systems. Adhikary told India Today Tech that he discovered the flaw while browsing the department’s website using scripts. He reported it about a week before the department validated and patched the vulnerability, crediting him on its acknowledgements page. Adhikary said he received no payment for the disclosure. He also reported a vulnerability in a US Department of Defense system, which authorities validated, with remediation still underway. Earlier this year, he gained attention after exposing digital security flaws in CBSE’s Online Submission of Marks portal. (PC: X)

Google will start sharing details of child sexual abuse material (CSAM) cases and repeat offenders with the Indian Cybercrime Coordination Centre (I4C) under the Ministry of Home Affairs, and a company spokesperson has confirmed the move. Until now, US-based platforms, including Google and Meta, reported such cases only to the US non-profit NCMEC, which then passed details to local law enforcement, often causing delays. Google said it has proposed putting the information-sharing into practice. Meta made a similar commitment after a BBC Eye investigation found Instagram running paid ads promoting child abuse material in India. Officials say more steps will be needed to strengthen child safety online. (PC: HT)

Three Indian-origin cybersecurity researchers used Anthropic's Claude to help break into OpenAI's systems during an authorised bug-bounty test in July, the Wall Street Journal reported. Harsh Jaiswal, Mohan Pedhapati and Rahul Maini of startup Hacktron AI found a flaw in OpenAI's community forum, then used Claude to help build a working exploit. They gained access to employee accounts and a route into a private GitHub environment. The test took under 72 hours and cost less than $3,000 in AI tokens. OpenAI fixed the flaws and paid a $6,500 bounty. The researchers, not Claude, linked the vulnerabilities together. (PC: India Today)