news aggregator

Researchers from Nanjing University and the University of Sydney developed an AI-powered bug-hunting agent that mimics human vulnerability discovery, validating flaws with proof-of-concept exploits. The Register reports: Ziyue Wang (Nanjing) and Liyi Zhou (Sydney) have expanded upon prior work dubbed A1, an AI agent that can develop exploits for cryptocurrency smart contracts, with A2, an AI agent capable of vulnerability discovery and validation in Android apps. They describe A2 in a preprint paper titled "Agentic Discovery and Validation of Android App Vulnerabilities." The authors claim that the A2 system achieves 78.3 percent coverage on the Ghera benchmark, surpassing static analyzers like APKHunt (30.0 percent). And they say that, when they used A2 on 169 production APKs, they found "104 true-positive zero-day vulnerabilities," 57 of which were self-validated via automatically generated proof-of-concept (PoC) exploits. One of these included a medium-severity flaw in an Android app with over 10 million installs.

Read more of this story at Slashdot.

Unavoidable AI has developers looking for alternative code hosting options

Among the software developers who use Microsoft's GitHub, the most popular community discussion in the past 12 months has been a request for a way to block Copilot, the company's AI service, from generating issues and pull requests in code repositories.…

An anonymous reader quotes a report from Deadline: Anthropic has agreed to pay at least $1.5 billion into a class action fund as part of a settlement of litigation brought by a group of book authors. The sum, disclosed in a court filing on Friday, "will be the largest publicly reported copyright recovery in history, larger than any other copyright class action settlement or any individual copyright case litigated to final judgment," the attorneys for the authors wrote. The settlement also includes a provision that releases Anthropic only for its conduct up the August 25, meaning that new claims could be filed over future conduct, according to the filing. Anthropic also has agreed to destroy the datasets used in its models. The settlement figure amounts to about $3,000 per class work, according to the filing. You can read the terms of Anthropic's copyright settlement here (PDF). A hearing in the case is scheduled for Sept. 8.

Read more of this story at Slashdot.

Whatever happened to that Baltra thing Tan and crew were helping Apple cook up?

Analysis OpenAI is allegedly developing a custom AI accelerator with the help of Broadcom in an apparent bid to curb its reliance on Nvidia and drive down the cost of its GPT family of models.…

Chair Carr calls E-Rate expansions unlawful, Ted Cruz warns of online risks for kids

The US Federal Communications Commission may soon pull funding for free Wi-Fi on school buses and in libraries after Chair Brendan Carr declared two Biden-era expansions unlawful and proposed eliminating them.…

A first-of-its-kind global research expedition has extracted freshwater samples from beneath the Atlantic Ocean floor off Cape Cod, documenting a massive aquifer stretching from New Jersey to Maine. The three-month Expedition 501, funded at $25 million by the National Science Foundation and European Consortium for Ocean Research Drilling, drilled up to 1,289 feet into the seabed at sites 20-30 miles offshore. Samples registered salinity as low as 1 part per thousand -- meeting U.S. freshwater standards -- with some readings even lower. Scientists collected nearly 50,000 liters for laboratory analysis to determine whether the water originates from ancient glacial melt or current terrestrial groundwater systems. The UN projects global freshwater demand will exceed supply by 40% within five years.

Read more of this story at Slashdot.

tldr; boffins did it

interview It all started as an idea for a research paper. …

Microsoft is giving away Microsoft 365 Personal subscriptions to all US college students. From a report: This subscription gives students free access to Microsoft's Office apps and the Copilot AI assistant integration for a year, after which the students are eligible for a 50 percent discount to continue the subscription. While most students have access to education versions of Microsoft 365 or Google Workspace, Microsoft's offer is for student's own personal Microsoft accounts, and is available to claim until October 31st. Microsoft 365 Personal is usually $99.99 a year, or $9.99 a month, and includes 1TB of OneDrive cloud storage.

Read more of this story at Slashdot.

Pro tip, don't install PowerShell commands without approval

A team of data thieves has doubled down by developing its CastleRAT malware in both Python and C variants. Both versions spread by tricking users into pasting malicious commands through a technique called ClickFix, which uses fake fixes and login prompts.…

Networks of pipes and heat exchangers can transfer excess heat from buildings into nearby bodies of water -- but as the world warms, the cooling potential of some water courses is now diminishing, Wired reports. Paris's district cooling network, which pipes Seine river water to cool 800 buildings including the Louvre Museum, faces diminishing returns as climate change warms water temperatures. The system achieves coefficients of performance between 4 and 15 -- significantly higher than conventional air conditioning -- by transferring building heat through heat exchangers to the river. The Seine briefly exceeded 27C this summer, approaching the 30C regulatory limit for returned water. The network currently spans 100 kilometers of pipes and will expand to 245 kilometers by 2042 to serve 3,000 buildings. Similar installations operate in Toronto using lake water from 83-meter depths and at Cornell University drawing 4C water from Lake Cayuga at 76 meters. Rotterdam and other cities are developing comparable systems as cooling demand rises.

Read more of this story at Slashdot.

After years of foot-dragging, penalties for blocking access finally kick in

It took four presidential administrations to finally get it done, but US health care actors that block patient and provider access to electronic medical data may finally begin to face actual consequences.…

An anonymous reader shares a report: Can AI help "smooth over" discussion on abortion, racism, immigration, or Israel-Palestine? Columbia University sure hopes so. The Verge has learned that the university recently began testing Sway, an AI debate program currently in beta. Developed by two researchers at Carnegie Mellon University, Sway matches up students with opposing views to chat one-on-one about hot-button issues and "facilitates better discussions between them," according to the tool's website. Nicholas DiBella, a postdoctoral scholar at CMU who helped develop Sway, told The Verge that about 3,000 students from more than 30 colleges and universities have used the tool. One of those may soon be Columbia. News of the potential partnership comes after more than two years of escalating tensions at Columbia between students, administrators, and the federal government. The university has spent years at the center of controversy after controversy: expulsions of pro-Palestinian student protesters, a string of police raids, and demands from the federal government. People at Columbia's Teachers College are testing Sway in order to potentially integrate it into the conflict resolution curriculum and "bridge-building initiatives at Columbia," DiBella said. He said there's also been interest from other teams at Columbia in using Sway for the fall 2026 semester and onward. Simon Cullen, an assistant professor at CMU and the other developer behind Sway, told The Verge that the company is also in touch with Columbia University Life.

Read more of this story at Slashdot.

9.9-rated flaw on the loose, so patch now

A critical code-injection bug in SAP S/4HANA that allows low-privileged attackers to take over your SAP system is being actively exploited, according to security researchers.…

Anthropic is blocking its services from Chinese-controlled companies, saying it's taking steps to prevent a US adversary from advancing in AI and threatening American national security. From a report: The San Francisco-based startup is widening existing restrictions on "authoritarian" regimes to cover any company that's majority-owned by entities from countries such as China. That includes their overseas operations, it said in a statement. Foreign-based subsidiaries could be used to access its technology and further military applications, the startup added. Anthropic's Dario Amodei has publicly advocated technological sanctions on China, particularly after DeepSeek stunned Silicon Valley with an advanced model this year. While Anthropic didn't name any companies, Chinese big tech firms from Alibaba to ByteDance have joined DeepSeek in an intensifying race to build AI services that can rival the likes of OpenAI in the US. Chinese entities "could use our capabilities to develop applications and services that ultimately serve adversarial military and intelligence services and broader authoritarian objectives," Anthropic said in its Friday post.

Read more of this story at Slashdot.

Much of the world favors protecting 30% of the world's land and water for nature by 2030, according to new research that has found overwhelming public support for the goal across eight countries on five continents. The Guardian: Nearly 200 nations agreed in 2022 to set aside 30% of the world's land and 30% of marine areas for nature. But just 17.6% of the world's land and 8.6% of the seas are now under global protection, and more than 100 nations are less than halfway to meeting the target, which was established under the 2022 Kunming-Montreal Global Biodiversity Framework. Governments will need to implement swift changes if they are to achieve the target within the next five years. But setting aside more space for nature can be a political pitfall. Often it can mean restricting people's access to land, halting resource extraction and relocating human settlements. These issues, along with possible effects on economic growth, are often cited by countries as barriers to expanding protecting areas. Research published last week in the Proceedings of the National Academy of Sciences, however, suggests that more than 80% of the public across eight sampled countries support the policy.

Read more of this story at Slashdot.

Because handing battlefield ID to an algorithm has never gone wrong before, right?

The US Army is preparing to deploy a new AI product that promises to automatically identify and track potential targets on the battlefield. However, humans will continue to make life and death decisions.…

Nobel laureate Geoffrey Hinton has warned that AI will concentrate wealth among a small elite while impoverishing most workers. The computer scientist, who pioneered neural network research in the 1980s, told Financial Times that rich people will use AI to replace workers, creating massive unemployment and profit increases. Hinton, who left Google in 2023 after selling his AI startup for $44 million a decade earlier, dismissed universal basic income as insufficient to address human dignity concerns from job losses. The 77-year-old physicist predicts superintelligent AI will arrive within five to twenty years. He blamed capitalism rather than AI technology itself for the coming economic disruption, stating the system ensures AI will primarily benefit the wealthy rather than solve grand problems like hunger or poverty.

Read more of this story at Slashdot.

It won't be fully functional for a while, though

video Europe's first exascale supercomputer has finally lived up to expectations, despite not being fully complete, as its general-purpose compute cluster is not set to be ready before next year at the earliest.…

Firefox 145 is dumping 32-bit Linux, though

Mozilla, the maker of Firefox, had some good news this week for users still clinging to Windows 7 – Firefox ESR 115 support is being extended until March 2026.…

Alphabet's Google was hit with a $3.45 billion EU antitrust fine on Friday for anti-competitive practices in its lucrative adtech business, marking its fourth penalty in its decade long fight with EU competition regulators. From a report: The move by the European Commission was triggered by a complaint from the European Publishers Council and comes amid a threat by U.S. President Donald Trump to retaliate against the European Union for any push against Big Tech. The EU competition enforcer had originally planned to hand out the fine on Monday but opposition from EU trade chief Maros Sefcovic on concerns about the impact on U.S. tariffs on European cars derailed EU antitrust chief Teresa Ribera's plan. The Commission said Google favored its own online display technology services to the detriment of rivals and online publishers and that it abused its market power since 2014 until today.

Read more of this story at Slashdot.