AI News Archive: June 17, 2026 — Part 16
Sourced from 500+ daily AI sources, scored by relevance.
- Singapore's May exports rise 38%, boosted by AI demand
Singapore's May exports rise 38%, boosted by AI demand Nikkei Asia
- Elon Musk's AI tool Grok was used in strikes against Iran: US govt
Elon Musk's AI tool Grok was used in strikes against Iran: US govt
- French President Macron to wind up G-7 summit with focus on AI, Trump dinner
French President Macron to wind up G-7 summit with focus on AI, Trump dinner The Straits Times
- AP Exclusive: Nvidia's Jensen Huang says society needs 'new social norms' in the age of AI
AP Exclusive: Nvidia's Jensen Huang says society needs 'new social norms' in the age of AI San Francisco Chronicle
- AP Exclusive: Nvidia's Jensen Huang says society needs 'new social norms' in the age of AI
AP Exclusive: Nvidia's Jensen Huang says society needs 'new social norms' in the age of AI Houston Chronicle
- Nvidia’s Jensen Huang shares 3 key points about the future of AI
Nvidia CEO Jensen Huang — whose work helped propel artificial intelligence — stressed in an Associated Press interview Tuesday that society needs to change with the advent of AI , arguing that a fuller embrace of the technology would improve people’s lives. Huang has been optimistic about AI’s potential to rapidly transform society, creating faster economic growth and more scientific breakthroughs. But as the head of a computer chip company now developing AI systems, he and others are confronting a public increasingly concerned about the potential harm the technology might bring. Huang has felt obligated to respond to critics who warn of job losses and threats to humanity itself. “We need to create new social norms,” Huang said in an interview. “I would advocate that everybody use AI. Just go engage it.” Huang made his case as AI has emerged as a political flashpoint, with objections to plans to build more data centers and fears that the speed with which it’s being adopted could spur the layoffs of workers who might not have a safety net. Such questions have threatened public support of the technology at a time when a race has kicked off with China , a contest Huang believes can best be won by a U.S. that is open to competing globally in AI. His close relationship with President Donald Trump also has been a source of criticism among Democrats, even as he emphasized that the computing power created by AI is vital to adding the factory jobs that have been promised for decades without much enduring success. It was an argument delivered by a 63-year-old man who has watched the technology develop and described himself as “boring” because his own life revolves mainly around work and his family. Huang disclosed during the interview some personal details, saying his favorite movie is “Kingdom of Heaven,” the 2005 epic about the 12th century Crusader Kingdom of Jerusalem. He said he had watched the movie “Project Hail Mary” three or four times and “I think we might watch it again this weekend.” Huang said the ability of AI to design a website, analyze complex documents, guide advanced research or even plan a kitchen remodeling has helped to close the technological divide in America. People can now do advanced work on computers without having to know how to program or write software, he added. Huang contended that there is a need for some government regulation and safety standards for AI, emphasizing that national security also needed to be a priority for the technology that has been powering stock market gains and U.S. economic growth in recent years. Huang said society will adapt to AI just as it did to automobiles. He said cars were once portrayed as killing children, but the world changed its norms by having sidewalks and crosswalks and stopping kids from playing in the streets. Huang skeptical of what government ownership of AI companies would achieve With a market capitalization of roughly $5 trillion, Nvidia has soared in valuation in recent years to become the world’s most valuable company. AI modeling companies OpenAI and Anthropic are potentially set to also clear the $1 trillion mark once their stocks are publicly traded. That explosive surge in wealth concentrated in AI companies has prompted renewed worries about economic inequality. Trump has tried to defuse those concerns, recently musing about the prospect that the U.S. government could own some shares in AI firms, so any windfalls would be more broadly shared with the public. That idea has also been advanced by Sen. Bernie Sanders, I-Vt., and even OpenAI CEO Sam Altman. Huang expressed skepticism about the idea, saying he expects the country will already benefit broadly from AI advancements. “I’m not exactly sure what they’re trying to achieve,” he said regarding government ownership. “I haven’t had a dialogue with them about that. But just remember that these are American companies. Their success benefits the stock price, of which many Americans are investors in. It generates taxes, which helps many Americans. It creates a lot of jobs.” He noted that AI companies could also lead to higher profits for energy, construction and hardware technology firms. “Americans have a stake in American companies already, naturally, in a whole lot of different ways,” Huang said. Huang says national security needs to be a priority on AI The Trump administration has recently reversed course from using a light touch on regulating AI to taking a heavier hand. It placed export controls on the AI company Anthropic’s latest models, leading the company on Friday to shutter all public access to those models over security concerns. Trump, a Republican, also signed an order to have new AI models voluntarily screened by the government before their release. Huang said the government was properly focused on national security issues, but it was important to provide clear guidance. “National security should always be the top concern of all technologies,” Huang said. “But having said that, you know, you have to be very specific about the risk that you’re concerned about, before setting up policies for export controls.” During the Biden administration, Nvidia pushed back against export controls that were designed to restrict its ability to sell chips to China, rejecting the administration’s premise that a ban would preserve an American edge on AI. Huang had warned that the export controls might limit America’s ability to develop the world’s AI ecosystem, as China would respond with its own advanced chips. Huang says energy is key problem for America’s AI development Huang stressed that the U.S. is vulnerable because of its deficient energy supply. The data centers performing the computations used in AI are creating a huge demand for electricity, which could be a strain on the power grid. Some data centers will be constructed with their own electricity sources, but Huang said the U.S. is starting from a disadvantage on energy. And without more energy, it can be harder to play to American strengths in its AI infrastructure, models and computer chip development. “The United States is woefully behind in energy production,” Huang said. “We just suffocated energy production for too long.” Huang complimented Trump on his approach to generating more energy in the U.S.. The president has aggressively supported the use of oil, coal and natural gas, but he has scorned the use of solar and wind power. The Nvidia CEO was not commenting on Trump’s opposition to climate-friendlier energy sources. But the gap he identified goes to some of the fears that U.S. households have about AI increasing their utility bills. Huang was speaking Tuesday in Sherman, Texas, at an expansion of the Coherent factory to develop a laser for transmitting data among chips, which could cut power use by AI systems by up to 50%. Trump’s fondness for Huang started at a Mar-a-Lago dinner Trump, not known for technological expertise, quickly developed a friendship with Huang. The president has called him “smart” and “amazing,” insisting that Huang accompany him on foreign trips. Most recently, Trump had Air Force One pick up the leather-jacketed CEO in Alaska while en route to his state visit to China. Their relationship started last year with an invitation to dinner at Mar-a-Lago, Trump’s home and private club in Florida. Huang was in the area to receive the Edison Achievement Award for his AI work. “He says drop by for dinner, and so I did,” Huang said. He went with his wife, Lori. “He was incredibly engaging, incredibly charismatic, conversational, asked a lot of questions,” Huang recalled. “From the moment that I met him, the only thing that he’s ever talked to me about is creating more jobs, reindustrializing the United States, protecting national security, winning.” He added that Trump “calls me in the middle of the night and wants to talk about one of these topics.” But his proximity to Trump has also led to criticism from Democratic lawmakers. Sen. Elizabeth Warren, D-Mass., objected to Huang not testifying before a Senate committee even as “he has time to attend a $1 million-a-head dinner at Mar-a-Lago.” Huang said he wants the U.S. president and other officials — regardless of party — to succeed. “We could differ with politics, but we should want him to succeed,” he said. “Because when President Trump succeeds, our country succeeds.” —Josh Boak, Associated Press
- AP Exclusive: Nvidia's Jensen Huang says society needs 'new social norms' in the age of AI
AP Exclusive: Nvidia's Jensen Huang says society needs 'new social norms' in the age of AI Austin American-Statesman
- French president urges US to share cutting-edge AI and democracies to cooperate on regulation
French president urges US to share cutting-edge AI and democracies to cooperate on regulation Boston Herald
- AP Exclusive: Bernie Sanders unveils plan to give the public direct ownership of AI companies
AP Exclusive: Bernie Sanders unveils plan to give the public direct ownership of AI companies Toronto Star
- AP Exclusive: Bernie Sanders unveils plan to give the public direct ownership of AI companies
AP Exclusive: Bernie Sanders unveils plan to give the public direct ownership of AI companies San Francisco Chronicle
- AP Exclusive: Bernie Sanders unveils plan to give the public direct ownership of AI companies
AP Exclusive: Bernie Sanders unveils plan to give the public direct ownership of AI companies Houston Chronicle
- Bernie Sanders unveils plan to give the public direct ownership of AI companies
Bernie Sanders unveils plan to give the public direct ownership of AI companies Boston Herald
- Banks face $170 billion profit squeeze while AI challengers reset finance
Banks face $170 billion profit squeeze while AI challengers reset finance Gulf News
- Half of AI job cuts will be reversed by 2027 — and it reveals the biggest mistake companies are making
Half of AI job cuts will be reversed by 2027 — and it reveals the biggest mistake companies are making Tom's Guide
- OpenAI launches scheduled tasks in ChatGPT, details here
OpenAI is updating ChatGPT with a new scheduled tasks feature. The new ability is rolling out starting today.
- Brain-inspired phototransistor could cut AI energy use by sensing and storing data
Inspired by the human brain, Oregon State University researchers have developed a new light-sensitive device that combines sensing and memory while controlling how digital memories strengthen or fade over time. The research was published in Advanced Functional Materials.
- Databricks targets AI-driven threat detection with Panther deal
Databricks targets AI-driven threat detection with Panther deal verdict.co.uk
- AMD acquires MEXT to boost AI-driven memory optimisation
US-based semiconductor firm AMD has announced the acquisition of MEXT, which specialises in AI-driven memory optimisation technology.
- Nvidia's Huang pledges AI will boost manufacturing jobs. A test will come in Texas
Nvidia's Huang pledges AI will boost manufacturing jobs. A test will come in Texas Austin American-Statesman
- Stop making boring slides because Google Vids just made AI avatars free for everyone
Google rolled out free AI avatars in Vids along with extended Veo video clips, multilingual voiceovers, and an upcoming emotion steering feature for personalized delivery.
- Google now lets you try its custom AI avatars without signing up for a paid plan
Your next Google Vids pitch could be in 24 languages without you saying a word.
- Millions of hours of Bay Area police body-camera footage go unwatched. Stanford says AI could change that
Millions of hours of Bay Area police body-camera footage go unwatched. Stanford says AI could change that East Bay Times
- UBP2: Uncertainty-Balanced Preference Planning for Efficient Preference-based Reinforcement Learning
Preference-based RL provides an approach to learning reward models from pairwise comparisons of behaviors, bypassing the need for explicit reward design. However, existing methods typically rely on passive data collection and suffer from poor sample efficiency, especially during the early stages of ...
- Rethinking Reward Supervision: Rubric-Conditioned Self-Distillation
Post-training of reasoning language models is commonly driven by supervised distillation and reinforcement learning with verifiable rewards. Distillation often relies on chain-of-thought annotations that are expensive to obtain and may themselves be noisy, incomplete, or partially incorrect; even wh...
- Explaining Attention with Program Synthesis
A longstanding goal of research on interpretable deep learning is to replace opaque neural computations with human-meaningful symbolic descriptions. In this paper, we propose an approach for approximating the behavior of components of deep networks with executable programs. We focus on attention hea...
- Trade-offs in Medical LLM Adaptation: An Empirical Study in French QA
The development of large language models (LLMs) has led to an increased focus on their adaptation to specialized domains and languages, yet the effectiveness of domain adaptation strategies remains unclear. We present a study of medical domain adaptation using French medical question-answering (QA) ...
- A Multi-Domain Benchmark for Detecting AI-Generated Text-Rich Images from GPT-Image-2
Text-rich images often contain privacy-sensitive, transactional, or decision-relevant information. As recent multimodal image generation models become increasingly capable of synthesizing realistic textual content and structured visual designs, detecting AI-generated text-rich images has become an i...
- X+Slides: Benchmarking Audience-Conditioned Slide Generation
Automatically generating slide decks from source documents is an important application of large language models (LLMs). Existing benchmarks primarily assess slide completeness and technical depth, while overlooking the target audience as a critical real-world factor. For instance, specialists demand...
- TxBench-PP: Analyzing AI Agent Performance on Small-Molecule Preclinical Pharmacology
Artificial intelligence (AI) agents promise to accelerate drug discovery by compressing interpretation and decision-making loops, but practical deployment requires trusted evaluation on realistic program decisions. We introduce TherapeuticsBench Preclinical Pharmacology (TxBench-PP), a verifiable be...
- STARE: Surprisal-Guided Token-Level Advantage Reweighting for Policy Entropy Stability
Reinforcement Learning with Verifiable Rewards algorithms like GRPO have emerged as the dominant post-training paradigm for complex reasoning in LLMs, yet commonly suffer from policy entropy collapse during training. We conduct a first-order gradient analysis of token-level entropy dynamics under GR...
- Machine Unlearning for the XGBoost Model with Network Intrusion Datasets
Machine Unlearning (MU) has emerged as an important technique for removing specific data points from trained models without requiring full retraining. However, most existing MU research focuses on deep learning and image data, leaving a gap in the domain of network intrusion detection, which relies ...
- Forecasting what Matters: Decision-Focused RL for Controlled EV Charging with Unknown Departure Times
The recent growth of EV adoption poses challenges for power systems, including increased peak demand and potential grid instability. Smart control of EV charging -- e.g., based on reinforcement learning (RL) -- can alleviate these issues by learning temporal and contextual patterns from historical d...
- Language Models as Interfaces, Not Oracles: A Hybrid LLM-ML System for Pediatric Appendicitis
Large language models (LLMs) can make clinical decision support more accessible by interpreting free-text documentation, but their direct use as diagnostic engines is limited by sensitivity to prompts, information order, and plausible but incorrect outputs. Structured machine-learning models offer m...
- User as Engram: Internalizing Per-User Memory as Local Parametric Edits
Personal memory in a language model is two problems: content and reasoning skill. The brain keeps the two apart (a sparse, local engram in the hippocampus for each episode, a slow neocortex for the shared skills that interpret it), so a new fact need not overwrite everything else. Most personalizati...
- Beyond Safe Data: Pretraining-Stage Alignment with Regular Safety Reflection
To achieve deeper safety alignment for large language models (LLMs), recent efforts have studied how to push safety interventions earlier into the pretraining stage, primarily by filtering unsafe data or rewriting it into safer forms. We argue that pretraining-stage alignment should go beyond making...
- A Technical Taxonomy of LLM Agent Communication Protocols
As large language models (LLMs) advance and multi-agent systems aim to overcome the limits of standalone agents, robust communication protocols are becoming essential infrastructure for distributed agent networks. Nonetheless, the fragmented protocol landscape presents a significant interoperability...
- Pareto Q-Learning with Reward Machines
We present Pareto Q-Learning with Reward Machines (PQLRM), a multi-objective reinforcement learning algorithm for tasks whose reward structure is specified by a set of reward machines (RMs). PQLRM combines Pareto Q-Learning (PQL), which maintains sets of vector-valued Q-estimates to approximate the ...
- Equivariant Graph Neural Networks Improve Optical Spectra Prediction for Materials Screening
Scalable prediction of optical spectra is a critical component of high-throughput materials screening for optoelectronic applications such as solar cells. Existing surrogate models are trained on spectra computed from lower levels of theory or rely on rotation-invariant scalar features, limiting the...
- Analysing drivers and interdependencies in European electricity markets using XAI
Electricity markets are inherently complex systems characterised by strong nonlinearities, high-dimensional interactions, and increasing interdependence across regions. While deep neural networks (DNNs) have demonstrated strong predictive capabilities for electricity prices, their lack of interpreta...
- Leadership as Coordination Control: Behavioral Signatures and the Recovery-Advantage Boundary in Multi-Agent LLM Teams
Team science holds that leadership is contingent: it helps only under specific conditions, and capable, autonomous teams may need none at all. We ask the analogous question for multi-agent LLM teams: under what measurable conditions does process-level coordination control add value, and do those con...
- ARIADNE: Agnostic Routing for Inference-time Adapter DyNamic sElection
The increasing deployment of parameter-efficient fine-tuning (PEFT) has led to model ecosystems in which a single backbone is paired with many task-specialized adapters. In this setting, inference-time queries often arrive without task labels, requiring the system to automatically select the most ap...
- Where Did the Variability Go? From Vibe Coding to Product Lines by Regeneration
In vibe coding, an emerging AI-driven paradigm, an LLM generates an entire program from a natural language prompt, but what happens to the variability that traditional software engineering carefully builds into code? To answer this question, we conducted an exploratory analysis on 10 vibe coded C/C+...
- A Hybrid LSTM--Vision Transformer Architecture for Predicting HRRR Forecast Errors
Forecast errors in high-resolution numerical weather prediction (NWP) systems are often linked to unresolved planetary boundary layer (PBL) processes, convection, terrain-induced circulations, and other vertically structured atmospheric phenomena. Previous work demonstrated that Long Short-Term Memo...
- FoMoE: Breaking the Full-Replica Barrier with a Federation of MoEs
Pre-training Large Language Models (LLMs) typically demands large-scale infrastructure with tightly coupled hardware accelerators. While increasing model and dataset scale remains the dominant driver of performance, Mixture-of-Experts (MoEs) architectures have recently achieved state-of-the-art resu...
- Spotlight: Synergizing Seed Exploration and Spot GPUs for DiT RL Post-Training
Reinforcement learning (RL) post-training of Diffusion Transformers (DiTs) is prohibitively expensive, requiring thousands of high-end GPUs. Existing works explore two directions to reduce cost: seed exploration improves training convergence by selecting high-contrast samples, yet adds compute to th...
- ThinkDeception: A Progressive Reinforcement Learning Framework for Interpretable Multimodal Deception Detection
Multimodal deception detection is critical for identifying fraudulent intentions, yet existing approaches predominantly rely on end to end black--box paradigms. These methods suffer from a severe lack of interpretability failing to provide transparent reasoning trajectories and struggling to explici...
- Beyond Tokenization: Direct Timestep Embedding and Contrastive Alignment for Time-Series Question Answering
Recent advances in large language models (LLMs) have given rise to time-series question answering (TSQA), which formulates time-series analysis as natural-language question answering. However, directly feeding raw numerical series into LLMs suffers from a tokenization bottleneck: Byte Pair Encoding ...
- CAPRA: Scaling Feedback on Software Architecture Deliverables with a Multi-Agent LLM System
Automated assessment in software engineering education has advanced significantly for code grading and essay scoring. However, reviewing software architecture deliverables, which requires analyzing structural completeness and requirements traceability, has not yet been fully automated. Applying Larg...
- A Controlled Benchmark of Quantum-Latent GAN Augmentation for Brain MRI
Medical image classification is often constrained by limited labeled data, motivating generative augmentation; recently, quantum generative models have been proposed for this purpose, frequently reporting accuracy gains. However, such claims are typically based on single training runs, do not match ...
- PaneFlow
Let AI agents build real animated slideshows