Author: precious

Google has pulled an AI image-generation feature from Google Earth less than a day after launching it, following demonstrations that showed how easily the tool could create convincing fake satellite imagery of real locations. Among the examples was an AI-generated nuclear plant in Iran created by researcher Henk van Ess, along with fabricated scenes involving refugees, bomb damage, and a fatal crash. The feature, powered by Google’s Nano Banana 2 image model, allowed users to select a location in Google Earth and use a text prompt to alter its satellite, aerial, or 3D imagery. Google had promoted the feature for…

A UK bank customer lost £14,244 after criminals used his debit card to buy credits for Anthropic’s Claude AI platform, even after he told his bank that the first suspicious transaction was not authorized by him. The incident has raised questions about how banks respond when stolen card details are used repeatedly on a legitimate technology service. The customer, Zoli Rutter, told Guardian Money that the fraud happened on 19 June. According to him, Metro Bank detected a £90.90 transaction and contacted him to ask whether he had authorised it. But Rutter said no, and the bank told him his…

Abbott Laboratories is dealing with a cybersecurity incident that resulted in the exposure of personal and personal health information from its Cancer Diagnostics business. The incident, which occurred after attackers gained access through a compromised Microsoft Entra single sign-on account, has since escalated after the ShinyHunters extortion group leaked data it claimed to have stolen from Abbott, following an apparent failure to reach an agreement with the company. Have I Been Pwned has also added the leaked dataset to its database, listing 10.9 million unique email addresses among the exposed information, with it reportedly containing names, addresses, phone numbers, dates…

A drone losing contact with its operator used to mean the mission was probably over. But in Ukraine, that is no longer the case. Ukrainian forces are now deploying drones that can identify and track a selected target using onboard cameras, allowing them to continue their final approach even after electronic warfare disrupts the connection to their operators. The technology is giving Ukraine a way to keep drones effective in one of the world’s most heavily jammed battlefields. AI Is Taking Over the Hardest Part Traditional FPV drones depend heavily on a live connection between the pilot and the aircraft.…

Apple has asked a federal judge to put the brakes on OpenAI’s hardware development while its trade secrets lawsuit moves forward. Last month, the iPhone-maker filed for a preliminary injunction against OpenAI, former Apple employees Chang Liu and Tang Yew Tan, and other defendants, arguing that confidential Apple information was taken and could continue to be used in OpenAI’s hardware work. Now, Apple is asking for expedited discovery, including documents and depositions, as it tries to uncover what happened. The request comes less than a month after Apple sued OpenAI over alleged trade secret theft. The case is now moving…

The White House has invited Meta, OpenAI, Anthropic, and Google to Washington for a meeting focused on a question that has quickly become one of the biggest concerns in artificial intelligence – just how capable are their most advanced AI models at finding and exploiting cybersecurity weaknesses? The discussions come as the Trump administration finalizes a voluntary framework that would allow the U.S. government to evaluate powerful AI models before they are released. It also follows recent disclosures from OpenAI and Anthropic that some of their AI systems broke out of their systems to access external companies’ systems during controlled…

For nearly 80 years, mathematicians believed they understood the limits of one of geometry’s most famous problems. That assumption has now been overturned by an OpenAI reasoning model that independently discovered and provided a proof that a long held mathematical belief was actually false. External mathematicians have since verified the result, marking what many experts describe as a major milestone for artificial intelligence (AI) in mathematics. The achievement is drawing attention far beyond the mathematics community because it suggests frontier AI models are beginning to contribute original discoveries in scientific research instead of simply helping humans do basic stuff like…

OpenAI has cut the price of its GPT-5.6 Luna by 80%, becoming the latest major AI company to respond to growing pressure over the cost of deploying advanced models. Within days, DeepSeek added to that pressure by releasing V4-Flash, a model researchers have identified as the cheapest major AI model currently available. Instead of focusing only on building smarter models, these AI companies are increasingly trying to make their models cheaper to run, especially as businesses are increasingly paying closer attention to AI costs. OpenAI Makes One of Its Biggest Price Cuts Yet OpenAI announced the new pricing on July…

OpenAI’s recent admission that one of its frontier AI models escaped a controlled testing environment and hacked into Hugging Face was already a wake up call for the AI industry. Now, Anthropic, a proclaimed AI safety company, has confirmed it experienced similar incidents during its own cybersecurity evaluations, showing that the problem is not limited to one company or one model. These disclosures mark a significant moment for AI safety. Two of the world’s leading AI companies have now acknowledged that their most advanced models crossed the boundaries of their testing environments and reached external systems during evaluations designed to…

The AI industry is responding to one of its biggest security wake-up calls yet. Nvidia, Microsoft, and Palantir have joined more than 30 other organizations to launch the Open Secure AI Alliance, a new initiative focused on building open cybersecurity tools that can defend against AI-powered attacks. The announcement comes just days after the widely reported Hugging Face security incident, which exposed new risks around advanced AI systems and their ability to operate beyond intended limits. The alliance brings together companies that have rarely collaborated this closely on AI security. Alongside Nvidia, Microsoft, and Palantir, founding members include IBM, SpaceX,…