AI & ML

What to expect from Google this week

Google is preparing to announce new AI developments at its I/O developer conference, with particular focus on addressing weaknesses in AI coding capabilities where competitors like Anthropic's Claude and OpenAI's models have gained significant advantages. The company is also expected to showcase advances in AI for science and health applications, areas where Google DeepMind has maintained leadership, including potential new scientific tools and the public release of its AI-powered Health Coach.

MIT Technology Review AI4h ago

Why Robots Are Suddenly Getting Really Good at Sports

Sony's table tennis robot just beat world-class human players. This isn't about ping-pong — it's about robots finally mastering the kind of split-second, unpredictable physical interactions that have stumped them for decades.

Creative Robotics4h agoOriginal

Inside Anduril and Meta's quest to make smart glasses for warfare

Anduril and Meta are developing augmented-reality military headsets that use AI, eye-tracking, and voice commands to help soldiers coordinate drone strikes and tactical decisions. The project, called Soldier Born Mission Command (SBMC), represents a $159 million Army contract to create glasses that overlay tactical information and use large language models (Google Gemini, Meta Llama, Anthropic Claude) to translate soldier speech into actionable military commands, with prototypes expected years away from deployment.

MIT Technology Review AI5h ago

Table tennis robot defeats some of world's best players - why this has major implications for robotics

Sony AI's table tennis robot Ace defeated some of the world's best players in recent evaluations, winning 3 out of 5 matches against elite competitive athletes, marking a significant advancement in AI systems operating in fast, uncertain real-world environments. The robot uses event-based vision sensors, high-speed cameras, and deep reinforcement learning to handle complex challenges like spin estimation and real-time ball tracking, demonstrating progress beyond controlled digital environments where AI has traditionally excelled.

Robohub11h ago

OpenAI and Dell partner to bring Codex to hybrid and on-premise enterprise environments

OpenAI and Dell have partnered to deploy Codex, OpenAI's AI coding model, in hybrid and on-premise enterprise environments, enabling organizations to use AI-assisted code generation securely within their own infrastructure. This partnership allows enterprises to integrate AI coding agents into their existing workflows while maintaining control over sensitive data and systems.

OpenAI Blog11h ago

AI reveals the invisible magnetic chaos wasting energy inside electric motors

Researchers at Tokyo University of Science have developed an AI-powered physics model called the entropy-feature-eXtended Ginzburg-Landau (eX-GL) model to understand and reduce energy loss in electric motors caused by complex magnetic domain structures. The model uses persistent homology and machine learning to analyze the chaotic 'maze-like' magnetic patterns in soft magnetic materials, revealing how temperature and microscopic structures trigger energy waste as heat, with potential applications for improving electric vehicle efficiency.

Science Daily Robotics17h ago

Your Robotaxi Is Being Driven by Someone in a Cubicle

Tesla's newly revealed crash data exposes an uncomfortable truth about autonomous vehicles: when things get tricky, humans take the wheel remotely. This isn't the self-driving future we were promised, and it raises serious questions about what 'autonomous' actually means.

Creative Robotics23h agoOriginal

Apple's new Siri app will reportedly offer auto-deleting chat options

Apple is reportedly planning to introduce auto-deleting chat features to its revamped Siri, allowing users to set automatic deletion of conversation logs after 30 days, one year, or never. The privacy-focused approach aligns with Apple's strategy to differentiate its AI from competitors by training on synthetic data rather than user chatlogs, positioning privacy protection as a core feature rather than an optional setting.

Engadget1d ago

Manufacturing Just Became the Most Interesting Robotics Story of 2025

While humanoids grab headlines and coding agents dominate the AI conversation, a quiet revolution is unfolding on factory floors. Mind Robotics' $400 million raise signals that the real money — and real innovation — is flowing toward general-purpose manufacturing automation.

Creative Robotics1d agoOriginal

Government AI Deals Are Moving Faster Than Government Can Handle

From Malta giving every citizen ChatGPT Plus to the UK handing its tax authority a £175 million AI contract, governments are racing to adopt AI at unprecedented speed. But the infrastructure, oversight, and public trust mechanisms needed to support these rollouts are lagging dangerously behind.

Creative Robotics2d agoOriginal

OpenAI is offering ChatGPT Plus to citizens of Malta for a year

OpenAI announced a partnership with Malta to provide free ChatGPT Plus subscriptions to all Maltese citizens for one year, making Malta the first country to launch such a large-scale government AI access program. Participants must complete an AI literacy course from the University of Malta before claiming their subscription, with the first phase launching in May 2026.

Engadget2d ago

YouTube's AI deepfake detection tool is now available to all creators 18 and older

YouTube has expanded access to its AI deepfake detection tool to all creators 18 and older, allowing them to identify and request removal of videos that use their likeness without authorization. The tool, which scans uploaded videos for facial matches and was previously limited to YouTube Partner Program members, journalists, and politicians, helps protect both creators and ordinary users from malicious or misleading AI-generated content.

Engadget2d ago

OpenAI and Malta partner to bring ChatGPT Plus to all citizens

OpenAI has partnered with Malta to provide ChatGPT Plus access to all citizens and offer training programs focused on practical AI skills and responsible AI use. The initiative aims to democratize access to advanced AI tools and build digital literacy across the population.

OpenAI Blog2d ago

Musk v. Altman week 3: Musk and Altman traded blows over each other's credibility. Now the jury will pick a side.

In the final week of the Musk v. Altman trial, lawyers traded blows over Elon Musk's and OpenAI CEO Sam Altman's credibility. Altman was grilled on his alleged history of lying and self-dealing involv...

MIT Technology Review AI2d ago

When AI Gets It Wrong, It Gives Out Your Phone Number

Recent reports show AI chatbots are leaking real phone numbers and personal data in their responses. This isn't a privacy crisis waiting to happen — it's already here, and it reveals fundamental problems with how these systems handle information they were never designed to protect.

Creative Robotics2d agoOriginal

New crash data highlights the slow progress of Tesla's robotaxis

Newly unredacted NHTSA data reveals that at least two Tesla robotaxi crashes since July 2025 occurred while remote teleoperators were controlling vehicles, including incidents where operators drove the vehicles into a fence and a construction barricade. The crashes, combined with reports of long wait times and navigation errors in Tesla's Austin and Dallas robotaxi services, highlight ongoing challenges in the company's autonomous driving development compared to competitors like Waymo.

Engadget3d ago

Further Notes on Our Recent Research on AI Delegation and Long-Horizon Reliability

Microsoft researchers present findings from their paper on AI delegation reliability, examining how language models preserve semantic content across long-horizon delegated workflows. The study finds that frontier models experience 19-34% degradation in artifact fidelity over 20 delegated iterations, though Python workflows show greater robustness, highlighting reliable long-horizon delegation as an important open research challenge.

Microsoft Research3d ago

AI Security Isn't a Feature Anymore — It's the Entire Product

From macOS exploits discovered by AI to chatbots leaking phone numbers, this week's news reveals a disturbing pattern: AI systems are simultaneously becoming our most powerful security tools and our biggest vulnerabilities. The industry is racing to patch holes while drilling new ones.

Creative Robotics3d agoOriginal

Making LLMs faster without sacrificing accuracy

Researchers present a new scaling law framework that relates architectural design choices in large language models to inference efficiency, demonstrating how to improve throughput by up to 47% without sacrificing accuracy. The work, presented at ICLR, identifies optimal configurations such as the MLP-to-attention ratio for models like LLaMA-3.2, addressing a gap in existing scaling laws that don't account for internal architectural decisions affecting inference performance.

Amazon Science3d ago

The Download: China's AI drama factory and the WHO's missing health targets

China's short drama industry is increasingly utilizing AI to generate content, with an average of 470 AI-generated dramas released daily in January, reducing production timelines from months to weeks and cutting costs by up to 90%. The article also covers several AI industry developments including Anthropic's $30 billion funding round at a $900 billion valuation, AI data centers straining the US power grid, and formal talks between Washington and Beijing on AI safety guardrails.

MIT Technology Review AI3d ago

The UK's tax authority is turning to AI to help identify fraud

The UK's tax authority HMRC has announced a £175 million, 10-year deal with AI firm Quantexa to use artificial intelligence for detecting fraud and tax return errors. The partnership will combine HMRC's data with other sources to identify fraudulent activities, unintentional errors, and legitimate payments made under wrong reference numbers, with all findings verified by human reviewers to ensure transparency and accountability.

Engadget3d ago

Security researchers, aided by Anthropic's Mythos, claim to have breached macOS

Security researchers used Anthropic's Claude Mythos Preview AI system to identify vulnerabilities and develop a privilege escalation exploit for macOS on M5 silicon chips. The work demonstrates how advanced AI systems can help discover previously unknown security bugs and attack paths, with Mythos part of Anthropic's Project Glasswing initiative aimed at using AI to prevent cyberattacks.

Engadget3d ago

NASA's new AI space chip could let spacecraft think for themselves

NASA is developing a radiation-hardened AI space chip that delivers up to 100 times more computing power than current spacecraft processors while surviving deep space conditions. The High Performance Spaceflight Computing processor, currently undergoing rigorous testing at NASA's Jet Propulsion Laboratory, could enable spacecraft to operate more autonomously and support faster scientific analysis for future missions to the Moon and Mars.

Science Daily Robotics3d ago

xAI introduces its coding agent called Grok Build

xAI has launched Grok Build, a coding agent designed to compete with products like Anthropic's Claude Code, initially available in beta to SuperGrok Heavy subscribers ($300/month). The release comes as xAI attempts to catch up with competitors in coding capabilities following founder Elon Musk's acknowledgment that the company had fallen behind on this front.

Engadget3d ago

Building a safe, effective sandbox to enable Codex on Windows

Learn how OpenAI built a secure sandbox for Codex on Windows, enabling safe, efficient coding agents with controlled file access and network restrictions.

OpenAI Blog3d ago

Databricks brings GPT-5.5 to enterprise agent workflows

Databricks announced that GPT-5.5 is now available for enterprise agent workflows, achieving state-of-the-art performance on the company's OfficeQA Pro benchmark with 50% accuracy and 46% error reduction compared to GPT-5.4. The model demonstrates significant improvements in parsing legacy documents, scanned PDFs, and orchestrating multi-step enterprise tasks through Databricks' AI Unity Gateway, AgentBricks, and Agent Supervisor API.

OpenAI Blog3d ago

Foundation Models Are Eating Robotics From the Inside Out

Mind Robotics just raised $400 million with foundation models at the core of its pitch. Microsoft is building foundation models for power grids. Material science is getting its own foundation model treatment. The robotics industry isn't just adopting AI—it's being reorganized around it.

Creative Robotics3d agoOriginal

Meta is bringing third-party apps and games to its display glasses

Meta is opening its Ray-Ban Display glasses to third-party developers, allowing them to build display-enabled apps and experiences optimized for the device's 20-degree field-of-view monocular display. The platform will support iOS/Android companion apps and web apps, with Meta suggesting use cases like information overlays, real-time data displays, streaming media, and minigames, though availability timeline remains unclear.

Engadget4d ago

Sea's View on the Future of Agentic Software Development with Codex

Sea Limited's Chief Product Officer discusses the company's deployment of Codex, an AI tool, across engineering teams to advance AI-native software development practices in Asia. The initiative represents a strategic adoption of AI-powered development tools to accelerate engineering productivity and innovation.

OpenAI Blog4d ago

OpenAI brings its Codex coding app to mobile

OpenAI has expanded its Codex coding assistant to mobile devices, allowing ChatGPT users on iOS and Android to access and control their Codex projects remotely through the ChatGPT app. The mobile integration uses a secure relay layer to keep credentials and files on local machines while enabling users to prompt Codex and receive updates like screenshots and test results while on the go, contributing to OpenAI's broader super app strategy.

Engadget4d ago

Netflix wants to use generative AI to make animated shorts

Netflix has launched a new studio called INKubator focused on creating generative AI-powered animated content, starting with short-form animations and potentially expanding to longer-form content. The studio is actively hiring key roles and represents Netflix's broader strategy of integrating AI into content creation, building on the company's existing use of AI in advertising and its acquisition of AI startup InterPositive.

Engadget4d ago

What Happens When Your Code Starts Writing Itself — And Nobody Knows Why

From OpenAI's secure coding sandboxes to automated prompt engineering, the latest wave of AI development tools promises to write better code faster. But as these systems become more autonomous and opaque, we're approaching a threshold where even experienced developers can't fully explain what their tools are doing—or guarantee what they'll do next.

Creative Robotics4d agoOriginal

Americans really don't want AI data centers close to their homes

A Gallup survey found that 71% of Americans oppose having AI data centers in their local areas, citing concerns about resource consumption (water, electricity), environmental pollution, and quality-of-life impacts. The poll highlights growing public resistance to data center expansion despite companies' need for increased compute power, with some jurisdictions implementing bans or moratoriums on new construction.

Engadget4d ago

The ChatGPT desktop app for Mac just got hit with a security breach

OpenAI found no evidence that user data was accessed.

Engadget4d ago

Mind Robotics raises $400M to scale AI-powered robots in manufacturing

Mind Robotics, spun out from Rivian by founder RJ Scaringe, raised $400 million to scale AI-powered general-purpose robots for manufacturing automation, bringing total funding to over $1 billion. The company is developing a full-stack platform combining foundation models, purpose-built robotics, and deployment infrastructure, with Rivian serving as its first customer and providing manufacturing environments for model training and data collection.

The Robot Report4d ago

Promptimus: Improving already good LLM prompts with zero manual engineering

Promptimus is an automated prompt-engineering framework that optimizes already well-developed LLM prompts without manual intervention by identifying specific failure points and suggesting targeted improvements. The method achieves best results on 16 of 20 benchmarks while outperforming six leading automatic prompt optimization approaches, demonstrating model-agnostic generalizability and sample efficiency across various enterprise tasks.

Amazon Science4d ago

Establishing AI and data sovereignty in the age of autonomous systems

When generative AI first moved from research labs into real-world business applications, enterprises made a tacit bargain: "Capability now, control later." Feed your proprietary data into third-party...

MIT Technology Review AI4d ago

Data readiness for agentic AI in financial services

This article examines the critical role of data quality, security, and accessibility in deploying agentic AI systems within financial services. It discusses how autonomous AI amplifies both the strengths and weaknesses of underlying data, and emphasizes the need for centralized, well-governed data infrastructure to meet regulatory requirements while maintaining accuracy and speed in rapidly shifting markets.

MIT Technology Review AI4d ago

Get a systems perspective to scaling humanoids at the Robotics Summit

The Robotics Summit & Expo in Boston (May 27-28) will feature a session on humanoid robotics scaling, led by Texas Instruments' Giovanni Campanella, covering system-level innovations in sensing, processing, power management, and edge AI for next-generation humanoid robots. The summit will include 50+ technical sessions and 70+ speakers from leading robotics and AI companies, addressing design challenges like sensor fusion, real-time processing, and motor control critical to humanoid performance.

The Robot Report4d ago

Anthro Energy partners with EnPower to develop lithium-ion cells in the U.S.

Anthro Energy and EnPower signed a memorandum of understanding to jointly develop and manufacture high-performance lithium-ion battery cells in the U.S. for robotics, autonomous systems, and defense applications. The partnership combines Anthro's advanced polymer electrolyte technology with EnPower's domestic manufacturing capabilities to create a vertically integrated U.S. battery supply chain, with plans to scale production beyond 750 MWh for applications including drones, unmanned vehicles, and embodied AI systems.

The Robot Report4d ago

Helping ChatGPT better recognize context in sensitive conversations

OpenAI announced safety updates to ChatGPT that improve its ability to recognize context in sensitive conversations, particularly around suicide, self-harm, and harm-to-others risks. The improvements use contextual cues across conversations and within individual chats to better distinguish between benign requests and those signaling potential harm, enabling more appropriate responses like de-escalation or crisis resource provision.

OpenAI Blog4d ago

Privacy Was Already Dead — Now AI Is Digging Up the Body

AI chatbots are exposing real phone numbers and personal information in their responses, surfacing data that was supposedly private. This isn't a bug — it's a feature of how these systems were built, and it reveals something uncomfortable about the training data powering the AI revolution.

Creative Robotics4d agoOriginal

Apple may open up the App Store to agentic AI

Apple is reportedly developing a framework to allow agentic AI services on the App Store while maintaining control over privacy, security, and preventing misuse. The company has previously blocked AI coding tools due to policy violations and revenue concerns, but faces pressure to participate in the growing agentic AI market among developers and consumers.

Engadget5d ago

AI chatbots are giving out people's real phone numbers

AI chatbots including Google's Gemini, ChatGPT, and Claude are exposing users' real phone numbers and personal information in their responses, either by surfacing actual contact details from training data or generating plausible-but-incorrect numbers that misdirect callers. Privacy experts warn this is a significant emerging risk with little recourse for affected individuals, and a privacy removal company reports a 400% increase in customer concerns about generative AI exposing personal data over the past seven months.

MIT Technology Review AI5d ago

Robot Hands Are Finally Getting the Attention They Deserve

While the robotics industry obsesses over legs and navigation, a quiet revolution is happening at the end of the arm. New foundation models and research breakthroughs are finally treating dexterity as a first-class problem — and it's about time.

Creative Robotics5d agoOriginal

GridSFM: A new, small foundation model for the electric grid

Microsoft introduces GridSFM, a lightweight foundation model that solves AC optimal power flow problems in milliseconds, enabling real-time optimization of electrical grids. The model uses a block-structured discrete neural operator to handle grids from 500 to 80,000 buses, potentially reducing up to $20B annually in congestion losses and unlocking 3.4 TWh of curtailed renewable energy.

Microsoft Research5d ago

Bridging the gap: Profitability meets sustainability through physical AI

Omar Asali, CEO of Ranpak, will present at the Robotics Summit & Expo on how physical AI and warehouse automation can simultaneously improve sustainability and profitability. The presentation will showcase real-world results including 25% reduction in corrugate usage, 75% decrease in box SKUs, and 4-5x throughput increases, while emphasizing human-centered automation that upskills workers rather than replacing them.

The Robot Report5d ago

Architecting Reliability: How GMSL Diagnostics Enable Robust Vision

This article discusses GMSL (Gigabit Multimedia Serial Link) diagnostics technology for autonomous and robotic vision systems, explaining how built-in diagnostic capabilities at the chip, link, and data layers enable early detection of sensor failures and signal degradation. The piece covers diagnostic requirements across the system lifecycle—from design and integration through production, runtime operation, and field service—to ensure robust and reliable perception systems in autonomous vehicles and robotics.

The Robot Report5d ago

Developing active and flexible microrobots

Researchers at Leiden University have developed microscopic flexible robots that move autonomously without sensors, software, or external control, instead relying on shape-environment interactions inspired by biological microswimmers. These 3D-printed chain-like microrobots demonstrate emergent intelligent behaviors including obstacle avoidance and object manipulation, with potential applications in targeted drug delivery and minimally invasive medical procedures.

Robohub5d ago

The Evolution of Vision Connectivity in Robotics: From USB and Ethernet to GMSL

This article examines the evolution of vision connectivity technologies in robotics, tracing the progression from USB and Ethernet-based camera interfaces to Gigabit Multimedia Serial Link (GMSL). The piece discusses how GMSL, originally developed for automotive applications, offers advantages like deterministic low-latency transport, better scalability for multi-camera systems, and improved mechanical robustness compared to legacy interfaces, making it better suited for modern robotics applications requiring real-time perception and sensor fusion.

The Robot Report5d ago