Author: precious

A company preparing to go public usually wants investors to see a steady hand at the wheel. So what should these investors make of two senior executives in a Big Tech leaving within the same week? That is the question hanging over OpenAI after longtime executive and COO Brad Lightcap and Chief Revenue Officer Denise Dresser announced their departures just days apart. The timing of their departures are particularly notable because OpenAI is already reorganizing its leadership as it prepares for a potential initial public offering (IPO). Two Senior Departures in Three Days Brad Lightcap, one of OpenAI’s longest-serving executives…

Texas spent years selling itself as the place where AI companies could build without too many obstacles. In fact, Governor Greg Abbott called the state the “epicenter of AI development” when Google announced a $40 billion investment in Texas in November 2025, with new data center campuses among the projects planned. But this attitude towards AI data centers seems to be changing quickly. In August, Abbott ordered Texas regulators to audit data centers seeking connections to the state’s main power grid before they can move forward. Electric Reliability Council of Texas (ERCOT) is now reviewing a queue containing more than…

How much work can an AI model actually do in cybersecurity now? OpenAI seems to answer that question, with its latest cybersecurity model having found flaws in one of the world’s most widely used browsers, Google Chrome. OpenAI’s GPT-5.6-Cyber has uncovered two previously unknown vulnerabilities in Chrome’s V8 JavaScript engine that could be chained to escape a security sandbox. This discovery gives a clearer picture of how far AI-assisted vulnerability research has advanced, while also showing why the same capabilities can create serious security risks when placed in the wrong hands. GPT-5.6-Cyber Found the Flaws in Chrome’s Engine OpenAI introduced…

More than 100,000 UK police officers and staff have had their names and work contact details published on the dark web after hackers breached the Police National Legal Database (PNLD), exposing information belonging to people who work in law enforcement and the wider criminal justice system. Detected in July, PNLD later confirmed that information including names, organisations, and work email addresses had been compromised and published online. However, the agency further said there is currently no evidence that passwords or other security credentials were accessed. This incident matters because the leaked information belongs to people whose jobs can make them…

Meta is facing a $942 million judgment in New Mexico after a state court found that its platforms contributed to harms affecting children. Ordering Meta to make changes to how Facebook and Instagram operate for users under 18, the judgment also combines a $375 million civil penalty imposed by a jury in March with another $567 million ordered by one Judge Bryan Biedscheid in August. The court treated the design and operation of Meta’s products as part of the child safety question, rejecting Meta’s argument that Section 230 protected it from liability for the products it designed. The Court Went…

For years, the idea of AI carrying out a cyberattack sounded like it was made up. A model could find vulnerabilities or write malicious code, but there was still a person deciding what happened next. But this became much harder in the advent of AI, and especially in July, when AI agents were used in a four-day operation against Taiwanese government systems and, according to researchers, kept the attack moving by finding new routes when earlier ones failed. Dream, the Israeli cybersecurity company that uncovered the operation, says the agents mapped 21 government systems, compromised at least 85 accounts, and…

Giving an AI access to a computer is one thing. But giving it the keys to an active cyberattack is another. In a campaign uncovered by Palo Alto Networks’ Unit 42, a threat actor did exactly that, allowing DeepSeek to investigate targets, find vulnerabilities, obtain exploit code and attempt attacks with almost no human input.. The operation was uncovered only after the agent accidentally exposed the attacker’s entire working environment, including API keys, exploit scripts, target lists, and attack logs. Unit 42 found that the attacker was using DeepSeek through the open-source Hermes Agent framework. The setup allowed the AI…

Manus is preparing to return to independent operations after Chinese regulators ordered Meta to reverse its acquisition of the AI agent startup, bringing an unusual end to a deal that was completed months earlier. Meta acquired Manus on December 29, 2025, in a transaction reported to be worth about $2 billion. However, in April, China’s National Development and Reform Commission prohibited the foreign acquisition and ordered the parties to withdraw from the transaction. The separation is now moving from a regulatory order to changes that users can see. Manus said it will resume operating independently, while certain users will have…

Nvidia has brought six of the world’s biggest investment firms into the AI infrastructure race, creating financing platforms designed to mobilize more than $500 billion in third-party capital. Partnered with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, the initiative is aimed at helping AI labs, cloud providers, and enterprises finance the expensive computing infrastructure needed to expand their operations. The scale matters because building AI infrastructure now requires far more than buying GPUs. Companies need servers, networking equipment, data centers and power capacity, which, in turn, creates a capital requirement that can stretch into billions of dollars before an…

Scientists have spent decades studying the genetic sequences that allow viruses to infect, replicate, and evolve. Stanford researchers have now used artificial intelligence to go a step further, designing complete viral genomes that did not previously exist and testing whether they could work. The team created 16 functional bacteriophages, viruses that infect bacteria, using genome language models developed with the Arc Institute. And the result marks the first reported demonstration that generative AI can design entire viral genomes that produce viable viruses. The research further gives scientists a new way to explore bacteriophages that could eventually help fight bacterial infections.…