AI News

Anthropic’s Mythos AI Reportedly Hacked the NSA’s Most Sensitive Systems ‘in Hours’

June 19, 2024 | Gizmodo AI
Anthropic’s Mythos AI Reportedly Hacked the NSA’s Most Sensitive Systems ‘in Hours’
When Anthropic first disclosed Mythos in April, it sent an anxious shockwave through much of the cybersecurity sector. The new AI model was allegedly so ruthlessly effective at finding and exploiting security vulnerabilities in existing software that the company said it was holding off on a public release and would only grant access to a small group of early testers, including the U.S. National Security Agency (NSA). Another wave of fear reverberated this

Researchers Identify 3 Key Drivers Behind 'AI Psychosis'

October 6, 2018 | Gizmodo AI
Researchers Identify 3 Key Drivers Behind 'AI Psychosis'
In just a year, AI psychosis became a term the internet is all too familiar with. Though some have contested the science behind it, various reports of psychotic breaks and, at times, fatal delusions linked to intense reliance on AI have flooded timelines. Some scientists now have a hypothesis on how chatbots could be directly responsible for these incidents. In a new study published in the Nature journal Digital Psychiatry and Neuroscience, researchers in

OpenAI Distances Itself From Nvidia With Jalapeño, Its First In-House AI Chip

September 4, 2021 | Gizmodo AI
OpenAI Distances Itself From Nvidia With Jalapeño, Its First In-House AI Chip
OpenAI fired the starting gun for the AI race in late 2022 with the launch of ChatGPT and has since become one of the most valuable startups in the world, reaching hundreds of millions of users every day. Its success, however, has depended on a web of partnerships with other companies that make the silicon chips upon which its AI systems depend. On Wednesday, OpenAI took its biggest step yet towards full technological independence with the unveiling of a n

Why drones and AI can't quickly find missing flood victims, yet

May 29, 2018 | TechXplore AI
Why drones and AI can't quickly find missing flood victims, yet
A machine learning algorithm identified piles of debris large enough to contain bodies in an aerial image of a flood aftermath. Credit: Center for Robot-Assisted Search and Rescue and University of Maryland For search and rescue, AI is not more accurate than humans, but it is far faster. Recent successes in applying computer vision and machine learning to drone imagery for rapidly determining building and road damage after hurricanes or shifting wildfire l

AI could make these common jobs more productive without sacrificing quality

November 8, 2023 | TechXplore AI
AI could make these common jobs more productive without sacrificing quality
Credit: Unsplash/CC0 Public Domain A study of Chile's workforce finds that AI could "accelerate" nearly half of the tasks performed by the country's 100 most common jobs. Generative AI has been pitched as a great leveler—a way to increase productivity, slash red tape and busy work, and narrow skills gaps. A new study by Gabriel Weintraub, a professor of operations, information, and technology at Stanford Graduate School of Business, quantifies the potentia

Indie game studios battle for piece of Switch 2 success

November 12, 2021 | TechXplore AI
Indie game studios battle for piece of Switch 2 success
At the BitSummit expo in Kyoto, indie designers touted their games. Players line up inside Nintendo's booth at a Japanese game fair to try upcoming Switch 2 titles, but, for once, the company's chirpy red-capped mascot Mario is nowhere to be seen. The three games on display are all made by smaller-scale, independent developers seeking to impress users of the device that became the world's fastest-selling console after its launch last month. But with Ninten

Democratizing AI-powered sentiment analysis

October 7, 2016 | TechXplore AI
Democratizing AI-powered sentiment analysis
Schematic of our hybrid sentiment analysis pipeline: transformer encoders generate fixed-length embeddings that feed into machine learning classifiers. Credit: Procedia Computer Science (2025). DOI: 10.1016/j.procs.2025.03.161 Artificial intelligence is accelerating at breakneck speed, with larger models dominating the scene—more parameters, more data, more power. But here is the real question: Do we really need bigger to be better? We challenged that assu

Scalable transformer accelerator enables on-device execution of large language models

March 25, 2026 | TechXplore AI
Scalable transformer accelerator enables on-device execution of large language models
Differences in processes with and without hardware accelerators. Credit: Electronics (2024). DOI: 10.3390/electronics13234683 Large language models (LLMs) like BERT and GPT are driving major advances in artificial intelligence, but their size and complexity typically require powerful servers and cloud infrastructure. Running these models directly on devices—without relying on external computation—has remained a difficult technical challenge. A research tea

A real-time look at how AI is reshaping work

January 3, 2018 | TechXplore AI
A real-time look at how AI is reshaping work
Credit: Unsplash/CC0 Public Domain Artificial intelligence may take over some tasks and transform others, but one thing is certain: it's reshaping the job market. Researchers at USC's Information Sciences Institute (ISI) analyzed LinkedIn job postings and AI-related patent filings to measure which jobs are most exposed, and where those changes are happening first. The project was led by ISI research assistant Eun Cheol Choi, working with students in a grad

New open-source tool makes complex data understandable

November 29, 2022 | TechXplore AI
New open-source tool makes complex data understandable
Screenshot of a Datavzrd report with annotated visual elements and controls. Credit: PLOS One (2025). DOI: 10.1371/journal.pone.0323079 Tabular data is at the heart of scientific analysis—whether in medicine, the social sciences, or even archaeology. Making it comprehensible and usable is often a tedious task, especially when the data is extensive or complex. The Institute for Artificial Intelligence in Medicine (IKIM) at the University of Duisburg-Essen h

Watermarks offer no defense against deepfakes, study suggests

August 8, 2018 | TechXplore AI
Watermarks offer no defense against deepfakes, study suggests
Images watermarked by StableSignature— non-semantic (top), and StegaStamp— semantic (bottom). The rightmost figures display the differences between the original and watermarked images that correspond to the changes encoding the watermarks. StableSignature's modifications are restricted to existing (high-frequency) edges such as wrinkles, hair, mustache, and intersections of multiple components. StegaStamp's watermark is distributed across the image. The ma

Trump's AI plan prioritizes deregulation to boost US dominance

March 19, 2019 | TechXplore AI
Trump's AI plan prioritizes deregulation to boost US dominance
Trump signed several executive orders to give his AI plan more legal weight, with his AI czar David Sacks (R) by his side. President Donald Trump unveiled an aggressive, low-regulation strategy on Wednesday to boost big tech's race to stay ahead of China on artificial intelligence and cement US dominance in the fast-expanding field. Trump's 25-page "America's AI Action Plan" outlines three aims: accelerating innovation, building infrastructure, and leading

Google-parent Alphabet earnings shine with help of AI

September 19, 2016 | TechXplore AI
Google-parent Alphabet earnings shine with help of AI
Google parent company Alphabet's cloud computing business is on pace to bring in $50 billion over the course of 2025, according to the tech giant. Google-parent Alphabet on Wednesday reported quarterly profits that topped expectations, saying artificial intelligence has boosted every part of its business. Alphabet's second-quarter profit of $28.2 billion—on $96.4 billion in revenue—came with word that the tech giant will spend $10 billion more than it prev

Sound insulation tiles at school may help calm crying children

January 19, 2024 | Phys.org AI
Sound insulation tiles at school may help calm crying children
A classroom in Japan with polyester fiberboard tiles on the ceiling to promote sound absorption. Credit: Ikuri Matsuoka When children are dropped off at a school or daycare for the first time, there can be a lot of feelings and sometimes meltdowns caused by being separated from parents, meeting new people, and hearing new noises. Could the architecture of the room help to soothe at least some of the children's concerns? "Classrooms without any sound absorp

Why media literacy matters—and how to stay well informed

January 13, 2022 | Phys.org AI
Why media literacy matters—and how to stay well informed

Cryo-em freezes the funk: Scientists visualize a pungent protein

July 10, 2023 | Phys.org AI
Cryo-em freezes the funk: Scientists visualize a pungent protein
Sulfite reductase is an enzyme that participates in sulfur metabolism. Until now, scientists have been unable to understand what the 3D structure of the enzyme looks like. Credit: Florida State University They say a picture is worth a thousand words. But it takes millions of pictures to understand the intricate chemistry of an enzyme that helps break down sulfur, commonly found in fruits, vegetables, alcohol and gasoline, into the colorless gas most noted

Phonon decoupling in naturally occurring mineral enables subatomic ferroelectric memory

May 31, 2025 | Phys.org AI
Phonon decoupling in naturally occurring mineral enables subatomic ferroelectric memory
Unique electric polarization of switchable BO4 tetrahedra in brownmillerite ABO2.5 a, Brownmillerite structure with alternating BO4 tetrahedral (Tet.) and BO6 octahedral (Oct.) layers viewed along the zone axis of [101]O. A site, B site and O ions are represented by white, dark gray and yellow spheres, respectively. Red and blue BO4 tetrahedra correspond to down- and up-polarized layers, respectively. Gray BO6 octahedra show centrosymmetric non-polar layer

Banks using AI are better at identifying creditworthy borrowers from afar, new study finds

October 2, 2022 | Phys.org AI
Banks using AI are better at identifying creditworthy borrowers from afar, new study finds

Unlocking the timecode of the Dead Sea Scrolls

February 24, 2020 | Phys.org AI
Unlocking the timecode of the Dead Sea Scrolls
For this study, writing styles in digitized manuscripts were analyzed using BiNet, a previously developed deep neural network for detection of handwritten ink-trace patterns. By combining writing styles with carbon-14 dates of manuscripts using artificial intelligence, the date-prediction model Enoch is able to produce an accurate date for the manuscript. Credit: Maruf Dhali, University of Groningen Since their discovery, the historically and biblically hu

A net a kilometer: Drones successfully identify ghost nets along NT coastline

December 14, 2022 | Phys.org AI
A net a kilometer: Drones successfully identify ghost nets along NT coastline