AI News Archive: April 28, 2026 — Part 18
Sourced from 500+ daily AI sources, scored by relevance.
- Protein Function Prediction with Pretrained ProtT5 Embeddings and Gradient Boosting
Protein function prediction remains a central challenge in computational biology due to the extreme sparsity and long-tail distribution of Gene Ontology (GO) [1] annotations. Advances in protein language models enable the extraction of dense, fixed-length representations from amino acid sequences, offering a scalable alternative to hand-picked features such as physicochemical properties. In this work, we evaluate a transformer-based embedding approach using ProtT5-XL combined with classical and modern multi-label classifiers for Gene Ontology prediction in the CAFA-6 setting. Fixed-length embeddings were generated via mean pooling of transformer hidden states and used as input to one-vs-rest logistic regression, gradient-boosted decision trees, and a neural network. Models were evaluated on held-out validation data with a focus on threshold selection, prediction sparsity, and behavior across frequent and rare GO terms. Gradient boosting consistently provided the best balance between pr
- Passive acoustic monitoring as a tool for early detection of invasive animal species: a systematic review and a case study
Invasive animal species are spreading rapidly across the globe, creating an urgent need for efficient early-detection and monitoring tools. Passive acoustic monitoring has become an established method in biodiversity research, but its application to invasive species monitoring has been less systematically explored. Here, we combine a systematic literature review with a field-based case study to evaluate the potential of passive acoustic monitoring for invasive animal detection. We identified 26 studies on acoustic monitoring of invasive animals, mainly addressing amphibians (11 studies), birds and fish (five each) with most studies from the USA and Australia. The use of acoustic monitoring of invasive species has increased during the past decade, with recent studies applying automated detection, machine learning, and large-scale monitoring frameworks. As a case study, we further tested the feasibility of low-cost acoustic monitoring of the invasive American bullfrog (Lithobates catesbe
- Topological attention asymmetry in ESM-2 attention implicitly encodes the allosteric hierarchy of the adenosine A2A receptor
Protein language models (PLMs) learn complex structural and functional dependencies from evolutionary sequence variation alone. While these models lack a temporal axis, it remains an open question whether their static attention maps encode the directional hierarchies characteristic of allosteric communication. We investigate this in the human adenosine A2A receptor using the ESM-2 transformer. We show that attention heads tuned to functional sites exhibit elevated structural asymmetry compared to the background. Using random-site and sequence-shuffle controls, we establish that this asymmetry is not an architectural artefact of the softmax operation, but a learned, sequence-dependent signal. By defining a signed pathway score between the extracellular ligand-binding triad and the intracellular G-protein interface, we identify a robust topological polarity: the model consistently routes information from the dynamic extracellular site to the conserved intracellular interface. This direct
- Distinct tasks engage a shared neural subspace in human hippocampus and anterior cingulate cortex
A hallmark of human cognitive flexibility is the ability to perform a wide range of unrelated tasks. While much is known about how neural circuits support individual tasks, less is understood about how these circuits support multiple tasks. Are neural representations largely task-specific, or do they retain a shared structure across distinct behaviors? To investigate this question, we recorded neural population activity from the hippocampus and anterior cingulate cortex, two regions linked to generalization and cognitive control, in eleven human patients performing three distinct tasks. Using dimensionality reduction, we estimated the neural subspace associated with each task and compared subspace geometry across tasks. We found that task-related subspaces were not independent: across tasks, approximately half of the subspace dimensions were shared, and these were primarily the dimensions containing most of the shared neural covariance. These findings indicate that neural population ac
- Detection of bronchopulmonary dysplasia in infants and prediction of school-age lung function from tidal breathing data using recurrent neural networks
Objective: To test whether machine learning (ML) models trained on tidal breathing flow time series can discriminate between individuals with and without respiratory disease and predict lung function indices obtained from conventional pulmonary function testing. Background: Accurate assessment of respiratory function in infants and young children is challenging because conventional pulmonary function testing requires sophisticated equipment and/or active patient cooperation. Tidal breathing measurements, in contrast, can be obtained non-invasively with little or no patient cooperation and at low cost, yet their clinical utility has been limited. We hypothesized that sufficiently long tidal breathing flow time series contain clinically relevant information that can be extracted using a recurrent neural network known as a long short-term memory (LSTM) network. Approach: We evaluated LSTM models in two scenarios within the Basel-Bern Infant Lung Development cohort. First, we assessed the
- Screening for patients at risk for cardiac amyloidosis via electronic health records: A multicenter machine learning development and validation study
Background Timely detection is crucial to improve outcomes in patients with cardiac amyloidosis (CA) by initiation of life-saving treatments. Although confirmatory bone scintigraphy is highly accurate for CA detection, identifying at-risk patients for referral remains challenging. Objectives This study aimed to develop and validate a machine learning model, Amylo-Detect, using structured multimodal electronic health record (EHR) data to guide referrals for confirmatory scintigraphy and monoclonal protein testing. Methods Consecutive all-comer patients (n=11,616) referred for bone scintigraphy at the Vienna General Hospital (2010-2023) were retrospectively included. Patients referred before August 2020 formed the development cohort. The remaining patients comprised the internal validation cohort. External validation was performed at the University Hospital Essen (n=1,521). Amylo-Detect was trained using 50 routinely available parameters to predict CA-suggestive uptake (Perugini grade >=
- Silent numerical failures in large language model-generated pharmacokinetic simulation code: a benchmark against target-controlled infusion validation criteria using the Marsh propofol model
Background. Large language models (LLMs) are increasingly used by clinicians to generate executable code for pharmacokinetic (PK) simulation. Whether such code meets the accuracy standards of target-controlled infusion systems has not been systematically evaluated. Methods. Five LLMs (ChatGPT, Claude, DeepSeek, Gemini, Grok) were prompted to generate Python code for the Marsh three-compartment propofol model under a standardized 120-minute bolus-plus-infusion regimen. Each LLM was tested in two phases: Phase 1, integrator free; Phase 2, fourth-order Runge-Kutta with 1-second step size mandated. Twenty runs per LLM per phase were collected (n = 200). Plasma concentrations were compared against a triple-validated reference using median prediction error (MDPE), median absolute prediction error (MDAPE), and Wobble. Runs were classified as Class A (MDAPE < 1 %), B (1-30 %), C ([≥] 30 %), or D (failed). Results. All 200 scripts were invokable and created a CSV file; 199/200 (99.5 %, 95 % CI
- A diagnostic model based on differential whole-brain dynamics for distinguishing neuropsychiatric symptom and cognitive impairment
Objectives: Neuropsychiatric symptoms (NPS) are prevalent in individuals of cognitive impairement (CI). However, the similarities and disparatenesses in whole-brain dynamics between individuals of CI and NPS are controversy. Electroencephalography (EEG) microstates reflect the whole-brain dynamics. This study aimed to investigate the differential EEG microstates parameters between CI and NPS and to construct related diagnostic model. Methods/design: This study was a cross-sectional study. Clinical and EEG data were collected, and an EEG microstate analysis were performed. The Least absolute shrinkage and selection operation (LASSO) regression model was used to identify significant differential EEG microstates parameters between CI and NPS and to construct a diagnostic model. The model performance was tested by the receiver operating characteristic curve (ROC). Results: This study enrolled 78 participants. A total of 36 EEG microstates parameters were identified and included in the diff
- Real-World Dose Modifications for FOLFIRINOX in Pancreatic Cancer: Evaluating the Feasibility of a Machine-Learning Framework
Background: FOLFIRINOX is a cornerstone regimen for eligible patients with pancreatic ductal adenocarcinoma (PDAC), but its clinical benefit is limited by substantial toxicity and frequent dose modification. In real-world practice, dose modifications are often individualized, and the clinical factors associated with these decisions remain incompletely characterized. Objective: To develop and evaluate an electronic medical record (EMR)-based machine-learning framework for modeling cycle-specific FOLFIRINOX dose modification decisions in patients with PDAC. Methods: We included patients with PDAC who received FOLFIRINOX at UCSF oncology clinics between November 2011 and December 2023. Predictors included demographic, clinical, laboratory, and treatment variables derived from the EMR. Logistic regression, random forest, and XGBoost models were trained using group-based 5-fold cross-validation to predict cycle-specific dose modifications for 5-fluorouracil, irinotecan, and oxaliplatin. Mod
- Neural and behavioural measures from attention testing show no support for efficacy of neurofeedback treatment for adult ADHD
Attention-deficit/hyperactivity disorder (ADHD) is associated with impairments in sustained attention and inhibitory control. Neurofeedback (NFB) is a widely used non-pharmacological treatment for ADHD and is generally well tolerated, but evidence for its efficacy remains mixed. Here we report results from secondary analysis of a randomized controlled trial of NFB training for adult ADHD, analysing behaviour and neural data from attention testing in both test-retest and treatment-vs-waiting list control group contrasts. We used electroencephalography (EEG) to investigate event-related cortical dynamics during the Test of Variables of Attention (TOVA), administered before and after NFB treatment. 44 adults with ADHD (NFB treatment, ADHD-T: n = 23; waitlist control, ADHD-W: n = 21) completed the TOVA before and after the NFB training period, while 128-channel EEG was recorded. Treatment-related change was examined through analyses based on behavioural TOVA performance, power spectral den
- Personalized, EEG-controlled intermittent theta burst stimulation
Brain-state-controlled transcranial magnetic stimulation (TMS) studies with real-time electroencephalography (EEG) show that the phase of ongoing oscillations modulates cortical susceptibility to TMS pulses. Translating this principle to repetitive clinical protocols, such as intermittent theta burst stimulation (iTBS), is an open challenge because within-train stimulation pulses corrupt real-time EEG. Moreover, the general difficulty of predicting EEG theta phase even to initiate an iTBS train applies. We present our solution for prefrontal EEG-phase-controlled iTBS, a personalized stimulation framework. We demonstrate the technical feasibility of aligning each train's initial bursts to the individual prefrontal theta phase and propose a "seed-and-sustain" hypothesis, whereby intra-train stimulation-induced entrainment at the individual theta rhythm carries the later bursts. Future human trials will be needed to evaluate the practical benefits of this approach.
- Resistivity-enhanced multi-physics machine learning framework for dynamic stress prediction in high sensitive UHPC
Resistivity-enhanced multi-physics machine learning framework for dynamic stress prediction in high sensitive UHPC EurekAlert!
- Microarchitecture Tailored to 3D-Stacked Near-Memory Processing LLM Decoding (U. of Edinburgh, Peking U., Cambridge et al.)
A new technical paper, “Rethinking Compute Substrates for 3D-Stacked Near-Memory LLM Decoding: Microarchitecture-Scheduling Co-Design,” was published by researchers at University of Edinburgh, Peking University, University of Cambridge, University of Chinese Academy of Sciences, and the Hong Kong University of Science and Technology. Abstract “Large language model (LLM) decoding is a major inference bottleneck because its... » read more The post Microarchitecture Tailored to 3D-Stacked Near-Memory Processing LLM Decoding (U. of Edinburgh, Peking U., Cambridge et al.) appeared first on Semiconductor Engineering .
- Here’s what OpenAI’s lawyer argued in his opening statement.
Here’s what OpenAI’s lawyer argued in his opening statement.
- Here’s what Musk’s lawyers argued in their opening statement.
Here’s what Musk’s lawyers argued in their opening statement.
- OpenAI Just Released GPT-5.5, and The Model is The Least Interesting Part
48 days. A super app and chief scientist who thinks the last two years were slow. GPT-5.4 launched in March. GPT-5.5 came out on April 23rd, 48 days later. Most coverage treated that gap as a footnote. It isn’t. It’s the entire story, wrapped up in a number most people read and immediately forget. What OpenAI announced last week wasn’t really a model release. It was a product strategy, an infrastructure bet, and a fairly blunt declaration of intent about what kind of company OpenAI wants to be. The model is the delivery mechanism. What it’s delivering is something else entirely. What GPT-5.5 actually does and what that actually means Yes, the benchmarks are real. GPT-5.5, codenamed “Spud” internally (charming, somehow), outperforms Gemini 3.1 Pro and Claude Opus 4.5 across standard evaluations. OpenAI grades its own homework, so take the exact margins with appropriate skepticism, but independent labs have corroborated the direction, and the gaps are wide enough that the directional sto
- Record $1.1B Seed Funding for Reinforcement Learning Startup
The vendor’s goal is achieving superintelligence.
- Amazon is already offering new OpenAI products on AWS
A day after OpenAI got Microsoft to agree to end exclusive rights, AWS announced a slate of OpenAI model offerings, including a new agent service.
- OpenAI loosens Microsoft ties, opens door to Amazon and Google Cloud
OpenAI loosens Microsoft ties, opens door to Amazon and Google Cloud
- OpenAI Partners with AWS, Breaking Microsoft Exclusivity
OpenAI partners with AWS, ending its exclusivity with Microsoft.
- Amazon now lets you have a real conversation with AI while shopping for products
Shopping on Amazon just got a lot more conversational. The company has launched Join the chat, a new interactive feature inside its existing Hear the highlights experience. If you have not come across Hear the highlights before, it is an AI-powered audio summary tool that lives on millions of product pages inside the Amazon Shopping […]
- Google expands Pentagon’s access to its AI after Anthropic’s refusal
After Anthropic refused to allow the DoD to use its AI for domestic mass surveillance and autonomous weapons, Google has signed a new contract with the department.
- Google signs classified AI deal with Pentagon, The Information reports
Google signs classified AI deal with Pentagon, The Information reports [Ads by RSSGenerator] Please try our other product: What is my IP address? [ One click Chrome ext ]
- Google Clears Pentagon to Use AI Tools in Classified Settings
Tech company added language to contract to say its AI wasn’t intended for domestic mass surveillance or autonomous weapons.
- Google Signs A.I. Deal With the Pentagon
The Pentagon has also signed deals for using A.I. on classified networks with OpenAI and Elon Musk’s xAI, amid a dispute with Anthropic.
- Google Signs Classified AI Deal With Pentagon Amid Employee Opposition
Google Signs Classified AI Deal With Pentagon Amid Employee Opposition The Information
- Google staff protest AI-Pentagon deal
Hundreds of staff signed a letter demanding Google’s CEO to refuse to allow the DoD to use its AI model for classified work.
- E.U. is pushing Google to give rival AI services the same Android access as Gemini
The European Commission says Google reserves key Android capabilities for Gemini and must give competing AI services equal access
- Google reportedly signs classified AI deal with US Pentagon
Tech company is latest Silicon Valley firm to sign agreement with US military despite widespread employee opposition Google has reportedly signed a deal with the US Pentagon to use its artificial intelligence models for classified work. The tech company joins a growing list of Silicon Valley firms inking agreements with the US military. The agreement allows the Pentagon to use Google’s AI for “any lawful government purpose”, the report from the Information added, putting it alongside OpenAI and Elon Musk’s xAI, which also have deals to supply AI models for classified use. Similar agreements, both at Google and other AI firms, have sparked significant disagreements with the Pentagon and major employee pushback. Continue reading...
- Google signs classified AI deal with Pentagon: Report
Google has joined other tech firms in a deal with the US Department of Defense. The agreement allows the Pentagon to use Google's AI for lawful government purposes. This includes sensitive work like mission planning and weapons targeting. The Pentagon is signing deals worth up to $200 million with major AI labs. Google's agreement includes adjustments to AI safety settings.
- Google signs AI deal with the Pentagon, ignoring protest from over 600 employees
Despite an open letter from hundreds of employees, Google has signed a contract giving the U.S. Department of Defense access to its AI models for classified work. Legal experts say the contract's safety clauses aren't legally binding. The article Google signs AI deal with the Pentagon, ignoring protest from over 600 employees appeared first on The Decoder .
- Google signed the Pentagon’s classified AI deal and walked away from its drone swarm contest on the same day.
Google has signed a deal allowing the Pentagon to use its Gemini AI models for classified military work under terms that permit “any lawful government purpose,” the company confirmed on Tuesday, one day after more than 580 Google employees signed a letter urging CEO Sundar Pichai to refuse exactly this kind of arrangement. The agreement provides the […] This story continues at The Next Web
- Google inks deal allowing Pentagon to use AI models for classified work
Google inks deal allowing Pentagon to use AI models for classified work The Japan Times
- Google allows Pentagon to use its AI in classified military work
Google allows Pentagon to use its AI in classified military work The Straits Times
- 600 Google staff urge CEO to reject classified US military AI contract
600 Google staff urge CEO to reject classified US military AI contract The Straits Times
- Move Over, Anthropic: Google Signs Deal to Supply Pentagon With Gemini AI for Classified Work
Anthropic and the Department of Defense had a blow-up over military use of AI. Now Google has stepped into the fray.
- Google Signs Pentagon AI Deal Despite Employee Backlash
The agreement reportedly allows the U.S. military to use Google’s AI models for classified work including 'any lawful government purpose.'
- Google and the Pentagon sign classified deal to give the Department of Defense unfettered access to its AI models
Google has signed a deal that allows the US Department of Defense to use its AI models for "any lawful government purpose." This is according to a report by The Information , which also notes that the full details of the contract are classified. An anonymous source within the company has suggested that the two entities have agreed that the search giant's AI tech shouldn't be used for domestic mass surveillance or autonomous weapons "without appropriate human oversight and control." However, the contract also reportedly doesn't give Google "any right to control or veto" anything the government decides to do. In other words, the famously trustworthy US government will just have to be taken at its word. “We believe that providing API access to our commercial models, including on Google infrastructure, with industry-standard practices and terms, represents a responsible approach to supporting national security,” a Google spokesperson told Reuters . The spokesperson also echoed that the com
- Google’s updated Pentagon deal uses Gemini for ‘any lawful government purpose’ with classified data
Amid opposition from employees, Google has signed a deal with the Pentagon which will allow the US government to use the company’s AI for “any lawful” purpose. more…
- Google and Pentagon reportedly agree on deal for ‘any lawful’ use of AI
The classified deal apparently doesn’t allow Google to veto how the government will use its AI models.
- Snapchat is rolling out sponsored AI agents
It was only a matter of time before they found a way to use AI agents as corporate shills. On Tuesday, Snapchat rolled out AI Sponsored Snaps, a "new way for brands to show up in Chat through AI agents." Or, put another way, it's conversational advertising. (Yay?) AI Sponsored Snaps will appear in the app's Chat tab (with a light gray "Ad" notation next to the brand name). After opening the chat, you can ask the agent questions about the brand it represents. Snap showed an example from its first partner for the initiative, Experian. The bot offers to answer your questions on saving money, improving your credit score and — there it is — exploring loans and credit cards. Whether through credit card offers or other means, the AI agent will presumably try to guide you toward behavior that makes money for the sponsor. So, it isn't clear why this would be better for consumers than asking a general-purpose chatbot like Gemini or Claude the same questions. Maybe the answer is as simple as, "It
- Google's "Ask YouTube" turns video search into a conversation
Google is testing "Ask YouTube," a conversational search feature that replaces the standard video list with a results page combining text, full-length videos, and Shorts. The article Google's "Ask YouTube" turns video search into a conversation appeared first on The Decoder .
- Microsoft's GitHub shifts to metered AI billing amid cost crisis
The all-you-can-eat AI buffet is coming to an end Microsoft is closing the AI buffet offered to GitHub Copilot customers, acknowledging that it can’t sell AI like Red Lobster's Endless Shrimp.…
- YouTube is testing an AI search mode that 'feels more like a conversation'
Google is determined to impose AI search onto as many of its products as possible, and the latest, er, victim is YouTube. A new feature called "Ask YouTube" will let you pose complex questions and receive "comprehensive results that include video and text, then ask follow ups to dive deeper," Google explained on its YouTube Labs page . The experimental feature is available starting today until June 8 for Premium US subscribers 18 and older. To use it, first, enable the feature in your account. Then, click on the new "Ask YouTube" button in the search bar and you'll see prompt suggestions, or you can enter your own, like "plan a 3-day road trip between San Francisco and Santa Barbara." After getting the results, you can try follow-up questions or choose from suggested prompts to explore in more detail. As shown in The Verge 's quick test , the prompt "short history of Apollo 11 moon landing" brought up a summary of the mission, along with videos and time stamps for relevant information.
- Is YouTube an AI Search Engine Now? Google Tests New In-App AI Chatbot
Is YouTube an AI Search Engine Now? Google Tests New In-App AI Chatbot PCMag
- YouTube Is Testing an AI Search Tool That Delivers Video and Text
YouTube Premium subscribers can try it out and shake up their video searches.
- YouTube is turning into an answer engine with a new conversational search feature
YouTube Premium subscribers in the US can now try Ask YouTube, a conversational AI search that pulls in videos, Shorts, and text summaries all in one place.
- YouTube is testing a chat-style search that cuts the scrolling
'Ask YouTube' replaces keyword hunting with back-and-forth questions.
- YouTube Tests AI-Powered 'Ask YouTube' Conversational Search Feature
YouTube is testing a new search feature that it says is meant to feel more like a conversation than a search interface. Users are able to ask complex questions in natural language, receive results that include video and text, and then ask follow-up questions. The new search option is part of YouTube Labs, an opt-in program that lets YouTube Premium subscribers gain early access to experimental features and prototypes. YouTube Labs is available to U.S. users, and subscribers sign up on the YouTube Labs website . According to YouTube, subscribers who opt in to try the new search can enter a prompt in the search bar, like "plan a 3-day road trip between San Francisco and Santa Barbara," and then select the Ask YouTube option to get the results. Search results include AI summary text, short videos, and long videos, with relevant segments in videos highlighted. The Ask YouTube search interface suggests some search prompts, such as "unique ideas for a backyard water feature," "plan a living
- Google is testing AI chatbot search for YouTube
‘Ask YouTube’ is a new way to search that generates an AI Mode-like page of information.