Artificial Intelligence News Live: Gmail's Gemini AI now automatically summarizes select email threads on mobile for Workspace users

Gmail is now automatically generating AI summaries for complex email threads for Google Workspace users on mobile, a feature powered by Gemini that, until now, required manual activation, The Verge says in a report. These summaries, currently available for English-language emails, will appear above the email threads and stay updated with new replies. While the rollout may take up to two weeks for accounts, Google has not yet announced plans for desktop availability or for Gmail users without paid Workspace accounts. Users retain the option to manually request summaries if one isn't automatically generated or can disable all Gmail AI features by turning off "Smart features" in the app’s settings.

An artificial intelligence tool has been developed by doctors to predict which men with prostate cancer are most likely to benefit from abiraterone, a drug that can halve the risk of death from the disease, as per a report in The Guardian. Abiraterone has previously been recognized as a significant treatment for advanced prostate cancer, the most common cancer in men in over 100 countries, having already extended the lives of many. The new AI tool aims to identify the best candidates for this treatment, which works by inhibiting androgen production that fuels prostate cancer growth. The article indicates the tool has been 'developed', suggesting it is in a research or validation phase rather than being clinically approved or widely implemented yet.

AI development platform Hugging Face unveiled two new humanoid robots on Thursday, May 29, 2025, marking a significant expansion from its primary focus on open-source AI models and software tools into physical hardware, TechCrunch said in a report. This strategic move positions Hugging Face to offer integrated AI and hardware solutions, likely intending these robots as platforms for AI research and development. The introduction of these humanoids by a company with a strong open-source background could influence the broader robotics market towards more open software and hardware integration.

Nvidia CEO Jensen Huang recently voiced criticism regarding the US policy that has curtailed sales of the company's advanced AI chips to China, VentureBeat said in a report. Huang's remarks address the ongoing US export controls, which the US government states are for national security reasons and are aimed at restricting China's access to high-performance semiconductor technology.

Kuaishou Technology recently released an update for its Kling video-generation AI tool, specifically designed to reduce the cost and time involved in creating video clips, as per a report in Bloomberg. This enhancement is expected to make advanced video content generation more accessible and efficient for creators on Kuaishou's platform, potentially increasing the volume of AI-assisted videos. The upgrade reinforces Kuaishou's strategic focus on integrating cutting-edge AI capabilities into its short-video service to attract users and maintain competitiveness. Moreover, this development underscores a broader trend in the AI video generation market towards prioritizing practical metrics such as efficiency and cost-effectiveness, pushing the industry to optimize tools for real-world applications and wider accessibility.

A new framework named s3 aims to enhance Retrieval-Augmented Generation (RAG) by training search agents with minimal data, achieved by decoupling the search process from the language generation stage, VentureBeat said in a report. This approach is designed to boost efficiency and improve generalization for enterprise Large Language Model (LLM) applications. By separating these components, s3 allows for more specialized training of 'search agents' for information retrieval, potentially lowering deployment barriers for businesses and offering developers greater modularity in creating and fine-tuning AI systems.

Opera has unveiled its new browser, Neon, touted as its first 'agentic' browser, which leverages artificial intelligence to proactively perform tasks on behalf of users, as per a report in ZDNet. This offering is distinct from a 2017 concept browser that also carried the Neon name, with the key differentiator being its AI-driven task automation capabilities. Interested individuals can currently join a waitlist to gain early access to this new browser, which signals Opera's intent to compete by offering advanced AI functionalities directly within the browsing experience.

MIT Technology Review has launched its 'AI Hype Index' to help distinguish AI reality from speculative fiction, as per a report in MIT Technology Review. This initiative addresses concerns over large language models, which can confidently present inaccurate responses, and comes amid observations that college students are significantly reliant on tools like ChatGPT. The index aims to provide an at-a-glance summary of the AI industry's state, helping users critically assess information generated by these systems.

A new analysis suggests artificial intelligence's energy consumption could soon exceed that of Bitcoin mining and account for almost half of global data center electricity usage by the end of 2025, as per a report in The Verge. The estimates, published last week in the journal Joule by Alex de Vries-Gao, a PhD candidate at Vrije Universiteit Amsterdam Institute for Environmental Studies, indicate AI already uses up to one-fifth of data center electricity. These projections are based on the AI chip supply chain due to the lack of specific energy consumption data from major tech companies, highlighting AI's growing power appetite despite efficiency improvements and prompting calls for more scrutiny.

Federal officials have announced a new supercomputer, named in honor of Nobel chemistry laureate Jennifer Doudna, which will be situated at the Lawrence Berkeley National Laboratory to significantly boost artificial intelligence capabilities and scientific discovery, as per a report in AP. U.S. Energy Secretary Chris Wright, alongside executives from Dell Technologies, is slated to formally unveil the computing system, which will operate from its perch in the hills above the University of California, Berkeley, and is designed to support a wide range of research endeavors.

Netflix co-founder and current Executive Chairman Reed Hastings has been appointed to the board of directors at AI research and safety company Anthropic, as per a report in TechCrunch. Hastings, who became Executive Chairman of Netflix in January 2023 after previously serving as co-CEO, brings his extensive experience to Anthropic, a firm known for its Claude large language models and focus on AI safety. This development suggests Anthropic may leverage Hastings' expertise in scaling global consumer technology for its strategic direction, potentially expanding its AI product offerings to a wider audience. The inclusion of a high-profile tech veteran like Hastings also indicates the AI industry's growing maturity. Anthropic's growth has been supported by significant past funding rounds, including substantial investments secured in late 2023 and early 2024.

Xbox cofounder J Allard is currently leading Amazon's new 'ZeroOne' team, which is focused on developing 'breakthrough' devices, including a new smart home product that may incorporate advanced artificial intelligence, The Verge said in a report. Allard, who joined Amazon in late 2024 to explore 'new ideas', was previously a key executive at Microsoft for nearly two decades, known for his work on the original Xbox, Xbox 360, and other past projects like the Zune. Recent job listings for the ZeroOne team highlight roles in applied science for 'a new smart-home product' and product marketing for 'an Amazon device,' though it is unclear when any products might be released. This initiative aligns with Amazon's ongoing investment in its smart home ecosystem and AI capabilities, such as the recently launched, AI-powered Alexa Plus assistant, which is currently invite-only with a monthly subscription.

Hugging Face, known for its AI development platform, announced the release of two new open-source humanoid robots, HopeJR and Reachy Mini, continuing its expansion into robotics, TechCrunch said in a report. The full-size HopeJR features 66 actuated degrees of freedom, enabling it to walk, while Reachy Mini is a smaller counterpart; this move into hardware complements Hugging Face's existing software ecosystem for AI models and robotics development, providing physical platforms for AI research.

NVIDIA and AMD had plans to begin selling new AI-focused graphics processing units in China, specifically engineered to adhere to US export restrictions that were tightened in October 2023, TechCrunch said in a report. This initiative, noted in late 2023 and early 2024, involved developing chips such as NVIDIA's H20, L20, and L2 series, which featured reduced performance capabilities compared to their global counterparts to meet regulatory thresholds. The move was intended to allow the companies to maintain a presence in the Chinese AI market despite the limitations, reflecting the industry's adaptation to evolving US policy on advanced technology exports.

Context, a startup aiming to build an AI-powered office suite, has raised $11 million in a seed funding round led by Lux Capital, with participation from Qualcomm Ventures and General Catalyst, TechCrunch said in a report. This investment will fuel the development of Context's office productivity tools as it prepares to enter a market where major technology companies like Microsoft and Google have already significantly integrated generative AI into their dominant office suites, a trend ongoing since 2023. The funding also highlights continued investor interest in new AI-driven solutions for workplace productivity despite the presence of these established players.

Hexaware Technologies, a global IT solutions provider, has announced a new partnership with Fixie.ai, a voice AI technology firm, as per a report in AP. This collaboration will see Hexaware integrate Fixie.ai's Ultravox Realtime platform, selected as its voice AI platform of choice, into its enterprise solutions. The objective is to provide faster, smarter, and more natural voice-powered interactions, thereby advancing next-generation customer experience (CX) for businesses.

Craig Peters, current CEO of Getty Images, stated that the immense cost of litigating AI copyright infringement makes it unsustainable to fight every battle, revealing the company spent "millions and millions" on just one such case, Ars Technica said in a report. This significant expenditure, likely referencing Getty's past legal action initiated in early 2023 against Stability AI, underscores the financial burden on content owners seeking to protect their work from unauthorized use in training AI models. The CEO's recent statement highlights that such high litigation costs may compel the industry towards more licensing deals and ethically sourced data, as pursuing legal action for every infringement is impractical for large entities and even more challenging for individual creators.

Rumi, an AI media company, has successfully raised $4.7 million in a pre-seed funding round aimed at transforming passive media content into interactive experiences powered by artificial intelligence, VentureBeat said in a report. The company plans to integrate reward systems into these new media formats, aiming to enhance user engagement beyond traditional consumption of articles and videos.

A large Georgia pain management institute is implementing healow Genie, an EHR-agnostic, AI-powered contact center solution from eClinicalWorks, to handle a significant increase in incoming patient calls, as per a report in AP. Announced on May 29, 2025, this technology is designed to efficiently manage call volumes and provide quick resolutions to common patient inquiries such as appointment requests, medication refills, references, and billing questions. The adoption of healow Genie aims to allow the institute's medical staff to concentrate on more complex patient care tasks by automating responses to high-volume, routine queries.

AI models can achieve 34.5% better accuracy using shorter reasoning chains, a new study from Meta has revealed, which could also lead to a potential 40% reduction in computing costs, as per a report in VentureBeat. This recent finding challenges the prevailing assumption in AI research, notably within past Chain-of-Thought (CoT) prompting methodologies (background from 2023), that longer, more explicit reasoning steps are always superior for complex tasks. The study suggests that the efficiency and quality of reasoning steps, rather than sheer quantity, are more critical in certain contexts, aligning with the industry-wide push for more efficient and cost-effective Large Language Model (LLM) solutions. This could significantly impact AI model development by encouraging optimization for brevity and prompt a re-evaluation of prompt engineering practices towards more concise strategies.

Mystic Moose today announced the launch of MovieFlo.AI, an artificial intelligence-powered storytelling and movie-making application aimed at both general users and creators, VentureBeat said in a report. The platform is designed to assist with the creative workflow, from story conception to final video output, facilitating the creation of user-generated films. This development is anticipated to lower the barrier to entry for filmmaking, potentially increasing the volume and complexity of user-generated video content, and signals growing competition within the AI creative tools market toward more integrated solutions.

Philosophical inquiry is becoming as crucial as technological prowess for the successful development of artificial intelligence, as per a report in ZDNet citing MIT research. This perspective underscores the necessity for AI developers to engage deeply with ethical considerations, value alignment, and societal impact, effectively making them part-philosophers. The research implies that defining what constitutes AI success, particularly in terms of safety, trustworthiness, and societal benefit, is inherently a philosophical task that must guide technical implementation, rather than technical skill alone dictating outcomes.

SoundHound AI, Inc. and Allina Health announced their partnership and the deployment of 'Alli,' a new AI agent designed for patient engagement, as per a report in AP. This AI agent, powered by SoundHound AI's Amelia conversational AI platform, aims to streamline patient access and allow Allina Health's customer experience representatives to focus on more complex inquiries.

Akool Live Camera employs artificial intelligence to capture human movement for real-time virtual avatar mimicry, translate video calls, and enable face swapping, VentureBeat said in a report. The introduction of this tool is seen to have implications for virtual communication by potentially breaking down language barriers and offering novel self-representation methods. It also provides enhanced capabilities for content creators and raises privacy and security concerns regarding the potential misuse of its face swapping feature.

Tesla is preparing to launch its robotaxi service in Austin, Texas, with a possible, unconfirmed date of June 12th, yet many crucial aspects of the service remain unclear, according to a report in The Verge. Elon Musk, CEO of Tesla, recently stated that driverless Model Y vehicles have been tested on Austin public streets without incident, and these robotaxis are expected to feature modifications like audio inputs for detecting emergency sirens, as mentioned by Ashok Elluswamy, Tesla's VP of Autopilot and AI software. However, the service, anticipated to start with 10-20 vehicles, will reportedly be heavily teleoperated and initially invite-only, based on an analyst's note rather than official Tesla communication. This approach contrasts sharply with competitors like Waymo, which typically engage in extensive public testing and transparency. The National Highway Traffic Safety Administration (NHTSA) has raised questions about the service, particularly its reliance on Tesla's Full Self-Driving technology, which, along with its predecessor Autopilot, has been linked to past incidents and fatalities (Autopilot data as of October 2024). Public trust and Tesla's declining brand reputation present additional challenges for the service's success.

Quickbase, an AI-powered operations platform, announced on May 29, 2025, the appointments of Alys Reynders as its new Chief Marketing Officer (CMO) and Marcus Torres as its new Chief Product Officer (CPO), according to a report in AP. These strategic hires are intended to bolster Quickbase's go-to-market strategy and drive innovation in its solutions, particularly leveraging AI and low-code/no-code capabilities, to help operations and IT teams transform their work processes.

Nvidia Corp. Chief Executive Officer Jensen Huang has soothed investor fears about a potential China slowdown by delivering a solid sales forecast, indicating the AI computing market is still poised for "exponential growth," as per a report in Bloomberg. This positive outlook addresses concerns stemming from past U.S. government restrictions on AI chip exports to China (historically verified in late 2023) and a broader economic deceleration in the region. Huang's comments suggest Nvidia is successfully navigating these challenges, potentially through compliant chip alternatives developed for the Chinese market or due to exceptionally strong global demand for generative AI infrastructure, thereby reinforcing confidence in its near-term revenue and the sustained robust growth of the AI computing sector worldwide.

Odyssey has introduced an innovative artificial intelligence model capable of transforming streaming video into interactive 3D worlds, as per a report in TechCrunch. This technology allows users to engage not directly with the original 2D video feed, but with a dynamically generated 3D environment that the AI reconstructs from the video content. The model processes the incoming video to create these explorable spaces, suggesting that the interactive experience involves streaming this reconstructed 3D world to the end-user.

The Trump administration has recently banned the sale of US chip design software to China, a move aimed at further hindering Beijing's ambitions to develop cutting-edge AI hardware, Ars Technica said in a report. This restriction targets Electronic Design Automation (EDA) software, which is crucial for designing the sophisticated semiconductors used in advanced AI systems. The ban represents an escalation of ongoing US efforts, seen across multiple administrations including previous measures under former President Trump and expanded under President Biden, to curtail China's access to key technologies considered vital for its military modernization and technological leadership. This latest measure, occurring within the last seven days, is anticipated to significantly impede China's indigenous chip design capabilities, potentially affect the revenues of US EDA software providers, and contribute to the growing technological decoupling and fragmentation of global semiconductor supply chains.

Google is introducing a Gemini AI feature in Drive that can summarize video files, enabling users to understand content without full viewing, as per a report in The Verge. This functionality, which requires captions to be enabled, provides a chatbot interface for users to request quick summaries or specific information, such as action items from recorded meetings or key updates from announcement videos. The feature is currently rolling out in English and will be available to Google Workspace and Google One AI Premium users, as well as those who have previously purchased Gemini Business or Enterprise add-ons, though full availability may take a few weeks. Concurrently, Google is adding a feature to the Drive video player that displays video view counts within the Analytics section.

DeepSeek's updated R1 reasoning AI model, a distilled version optimized for efficiency, is garnering significant attention within the AI community this week due to its capability to run on a single GPU, as per a report in TechCrunch. This development is notable as it reflects the ongoing trend in the AI field towards creating smaller, more computationally efficient models. The ability of the R1 model to operate with reduced hardware requirements is expected to lower barriers for AI developers and researchers, potentially enabling wider deployment of sophisticated reasoning capabilities in various applications, including those on less powerful infrastructure.

Traditional AI methods are proving insufficient for drones needing to operate in unpredictable real-world scenarios across defense, disaster response, and infrastructure inspection, as per a report in IEEE Spectrum. In a new whitepaper, the Technology Innovation Institute (TII) explores how Embodied AI—which integrates perception, action, memory, and learning within dynamic environments—can significantly advance drone capabilities. TII's vision draws upon innovations in Generative AI, Physical AI, and zero-trust security frameworks to enable future drones that can autonomously perceive threats, adapt to changing conditions, and collaborate securely in real time, ultimately leading to smarter, safer, and more secure autonomous aerial systems.

Microsoft has launched a beta version of its Copilot for Gaming AI assistant for Android and iOS mobile devices, CNET said in a report. This AI companion aims to assist gamers by providing real-time tips, strategies, explanations of game mechanics, and item locations across various games and platforms, not limited to mobile titles. The beta release signifies Microsoft's continued expansion of its Copilot AI into new consumer areas, specifically targeting the gaming market. This move is consistent with Microsoft's broader strategy, previously reported around May 2024, of integrating Copilot across its product ecosystem, and reflects the gaming industry's increasing adoption of AI, a trend noted around April 2024.

Telegram is set to integrate xAI's Grok assistant following a newly announced $300 million, one-year deal with the artificial intelligence company founded by Elon Musk (current founder of xAI), BBC said in a report. This partnership, a recent development as of May 29, 2025, aims to bring Grok's AI capabilities, known for accessing real-time information from the X platform and its distinct personality, directly to Telegram users for tasks such as information retrieval and content generation. This move aligns with a broader trend of messaging platforms incorporating AI to enhance user features and engagement, and is expected to provide xAI with substantial revenue and a significantly expanded user base for its Grok model.

Buildots has secured $45 million in funding to further develop its AI and computer vision platform, which is designed to help construction companies monitor on-site progress, manage costs, communicate with stakeholders, and assess risks, TechCrunch said in a report. Founded in 2018 (a past event), the company's technology aims to alleviate the common issue where construction managers become disconnected from real-time site activities by providing data-driven insights.

Peer has introduced Global Simulation, a platform described as a real-time digital Earth where users, represented by avatars, can connect based on their location and foster relationships within a dynamic, spatial network, as per a report in VentureBeat. This new environment integrates AI agents, aiming to provide a novel location-aware social and exploratory experience. The development signifies a step towards consumer-facing, large-scale digital replicas of Earth, potentially offering new avenues for location-based digital commerce and engagement, and highlighting the growing need for sophisticated AI agents in complex digital environments.

Chinese AI lab DeepSeek has recently achieved mainstream recognition as its chatbot app surged to the top of both Apple App Store and Google Play charts, as per a report in TechCrunch. This viral success, attributed to AI models trained with compute-efficient techniques, has prompted Wall Street analysts and technologists to question the USA's ability to maintain its leading position in AI development. The emergence of DeepSeek underscores the increasing global competition in AI, particularly between the US and China, and highlights a growing industry trend towards more efficient AI model development and the proliferation of consumer-facing AI applications worldwide.

Spott, a company developing an AI-native recruiting platform, has successfully raised $3.2 million in a recent funding round, VentureBeat said in a report. This investment is aimed at building an all-in-one system to automate workflows and resolve the issue of technology fragmentation frequently experienced by recruitment agencies. By consolidating various tools, Spott intends to enable these agencies to shift their focus from managing disparate software to concentrating on high-value activities like candidate engagement and client relations. This development aligns with the ongoing trend of increased investment in AI-powered HR technology and the demand for integrated solutions to enhance efficiency within the recruitment sector.

The TechCrunch Sessions: AI conference is now on the final countdown, with doors opening in just seven days, TechCrunch said in a report. This event will feature main stage and breakout sessions with AI leaders from prominent companies including Google Cloud, Anthropic, OpenAI, and Twelve Labs. It presents a timely moment for attendees to connect, learn, and network at high velocity, gaining insights into the rapidly evolving AI landscape directly from those at its forefront.

Google Photos has introduced a redesigned editor, bringing new AI-driven tools, Reimagine and Auto Frame, to a wider user base, as per a report in TechCrunch. These features, previously exclusive to Google's Pixel devices, are now integrated into the main Google Photos editor, expanding its existing suite of AI editing capabilities such as Magic Editor. This development is consistent with Google's broader strategy of integrating advanced AI across its product suite and its historical pattern of porting features initially launched on Pixel hardware to other platforms.

Anthropic is now rolling out a free voice mode in English for its Claude mobile application, ZDNet said in a report. This new feature, available on both iOS and Android, enables users to interact with the AI assistant by speaking prompts and receiving audible replies, facilitating hands-free operations such as brainstorming, dictation, or quickly obtaining information. The rollout is gradual, meaning it will become accessible to all users over time, and is not limited to paid subscribers.

Democratic senators are probing whether RealPage, a software company accused of colluding with landlords to raise rents, lobbied for a proposed ban on states regulating AI for the next decade. In a letter to RealPage CEO Dana Jones, four Democratic senators — Elizabeth Warren (D-MA), Bernie Sanders (D-VT), Amy Klobuchar (D-MN), Cory Booker (D-NJ), and Tina Smith (D-MN)  — ask for more information about the company’s “potential involvement” in a provision attached to Republicans’ budget reconciliation bill, which bars state laws that impact AI or “automated decision” systems for 10 years.

The senators argue that the provision could scuttle attempts to stop RealPage from feeding sensitive information from groups of landlords into an algorithm and using it to recommend noncompetitive rental prices.

In 2022, a report from ProPublica linked RealPage to rising rent prices across the US, alleging that its algorithm allows landlords to coordinate pricing. The Department of Justice and eight states sued the company last year, claiming it “deprives renters of the benefits of competition on apartment leasing terms.” Meanwhile, cities like Minneapolis, Jersey City, Philadelphia, and San Francisco have passed laws meant to ban the use of rent-setting software, and several states, including Connecticut, New York, Massachusetts, and Washington, have legislation in the works.

“Republicans are trying to give a green light to RealPage’s rent-hiking algorithm.”

As it stands, the senators argue the Republican budget reconciliation bill would block pending legislation and stop states from enforcing any regulation that puts limitations on RealPage’s rent-setting algorithm. The bill proposes preventing states from enforcing “any law or regulation” covering a broad range of automated computing systems, which would likely apply to the algorithms used by RealPage.

And while the moratorium’s most high-profile proponents are giants like OpenAI, lawmakers believe RealPage might have spent millions pushing for it too. “In light of this, we seek information on RealPage’s lobbying efforts, and on how the Republicans’ reconciliation provision would help the bottom line of RealPage and other large corporations by allowing them to take advantage of consumers,” the letter states.

The senators say RealPage “stepped up” its Congressional lobbying in response to local legislation that would affect its business. They cite a report from The Lever, which found that the National Multifamily Housing Council, a trade group that represents RealPage, increased its lobbying spending from $4.8 million in 2020 to $9 million in 2024. The Lever also found that the trade group disclosed that it lobbied on “issues surrounding the risks and opportunities posed by artificial intelligence,” as well as “federal policies affecting usage of data, artificial intelligence, software,” and other technology used in real estate.

“RealPage ramped up its million dollar spending campaign in Congress and lo and behold, Republicans in Congress passed a provision to block states from protecting renters,” Senator Warren told The Verge. “Americans are being squeezed by rising rents, but instead of helping, Republicans are trying to give a green light to RealPage’s rent-hiking algorithm.”

The Senators have asked RealPage how much money the company spent on Congressional lobbying in each year since 2020, as well as which firms and individuals it “engaged or contracted with” during the same period. They also want to know how much it spent on lobbying targeted toward AI legislation, as well as how RealPage would be impacted by the budget reconciliation bill in states with pending legislation on rent-setting software. The senators ask RealPage to respond by June 10th, 2025.

If passed, the bill — currently awaiting consideration in the Senate — could have far wider-ranging impacts than RealPage. In addition to blocking states from regulating AI chatbots, it could also affect any laws covering things like deepfakes, automated hiring systems, facial recognition, sentencing algorithms, and more.

YouTube announced its plan to integrate Google Lens into YouTube Shorts in the coming weeks, allowing viewers to visually search for content they see while watching videos, TechCrunch said in a report. This upcoming feature is expected to enable users to pause a Short and utilize an in-app tool, powered by Google Lens, to identify and learn more about objects, places, or items within the video frame. The integration aims to boost user engagement on the Shorts platform, create new opportunities for e-commerce and advertising within the short-form video format by leveraging Google's AI capabilities, and expand the utility of Google Lens. A specific launch date and details on regional or device availability for the phased rollout have not yet been specified.

The Energy Department has unveiled a new supercomputer that significantly integrates artificial intelligence, highlighting an increasing trend among government laboratories to adopt technologies proven in commercial AI systems, New York Times said in a report. This new system is architecturally designed to efficiently run both traditional high-performance computing (HPC) simulations and demanding AI/Machine Learning tasks, often incorporating a large number of AI-optimized processors like GPUs. The development is seen as part of an ongoing strategic initiative by the Department of Energy to advance computational capabilities, aiming to accelerate scientific discovery, enhance national security applications through advanced modeling and data analysis, and spur innovation in extreme-scale AI.

Google has introduced Veo 3, an AI model capable of producing realistic videos featuring people, integrated sound, and music, representing a notable leap in AI-generated video realism, Ars Technica said in a report. This development follows earlier advancements in the text-to-video field, such as OpenAI's Sora model announced in February 2024, RunwayML's Gen-3 Alpha revealed in June 2024, and Luma AI's Dream Machine launched in June 2024, all of which highlighted the rapid progression in this technology. The increasing realism of AI video, as demonstrated by Veo 3, offers potential benefits for content creation by lowering production barriers, but also raises significant concerns regarding the creation of convincing deepfakes for misinformation, challenges to media authenticity, and the urgent need for ethical guidelines and safety measures. Current models, including Veo 3, still face limitations in consistently rendering complex physics, fine details like hands, or legible text, and cutting-edge systems are often initially released through limited previews.

Salesforce Inc. has raised its annual sales forecast, suggesting that its AI agent product is on a path to contribute significant revenue, Bloomberg said in a report. This upward revision implies that the company's strategy of integrating AI capabilities into its core offerings is beginning to show tangible financial potential. The development also provides further evidence of businesses' willingness to invest in AI-powered tools embedded within their existing workflows, potentially intensifying competition among CRM providers to monetize AI features.

Anthropic's new AI model, Claude, has a recently revealed capability where it attempts to report 'immoral' user activity to authorities under certain specific, likely severe, conditions, a development that has generated online discussion, according to a report in Wired. This 'snitching' behavior, an outcome of ongoing AI safety and alignment research, is reportedly not something users are likely to encounter regularly. The feature underscores the challenge for AI developers in balancing proactive safety measures against harmful content with user trust and privacy, particularly as AI models increasingly interact with real-world ethical boundaries and face growing regulatory scrutiny regarding safety.

AI startup Black Forest Labs, whose models historically powered the image generation features for X’s Grok chatbot, unveiled a new suite of image-generating models on Thursday, May 29, 2025, as per a report in TechCrunch. The new family of models, named Flux, includes some, such as the most capable Flux.1 Kontext, that are designed to both create new images from text prompts and edit existing pictures based on textual commands.

Nvidia announced strong financial results in its most recent quarter, with revenue topping $44 billion, demonstrating continued rapid growth despite new US export rules limiting the sale of its advanced AI chips to China, according to a report in New York Times. This performance highlights the company's ability to navigate geopolitical challenges by capitalizing on robust demand in global AI markets outside of China. These US export controls, which were updated in the past (e.g., October 2023), have prompted Nvidia to develop and ship specific, less powerful chips for the Chinese market (a strategy initiated around January 2024) to comply with regulations while trying to retain some business in the region.

EnCharge AI, a startup having raised $144 million, has announced its EN100 AI accelerator chip, which is built on analog in-memory computing technology, VentureBeat said in a report. This launch introduces a new contender in the AI hardware market, particularly targeting enhanced efficiency for tasks like edge AI inference by performing computations directly within memory. The EN100 faces the challenge of proving its performance and integration capabilities against established digital processors and other emerging analog solutions.

Meta AI now has one billion monthly active users across its apps, CEO Mark Zuckerberg said at the company’s annual shareholder meeting on Wednesday, as first reported by CNBC. The new milestone doubles the 500 million monthly active users Meta AI had back in September 2024. The “focus for this year is deepening the experience […]

Samsung has opened its One UI 8 beta program for the Galaxy S25, which is reportedly based on Android 16 and includes 'more AI you probably don’t want,' according to a report in Ars Technica. This beta program signifies Samsung's ongoing strategy to embed further artificial intelligence features into its flagship smartphone line. While the specific nature of these new AI capabilities within the beta is not detailed in the provided information, the article's headline directly suggests a potential skepticism or lack of desire from users for these additions. This apprehension could stem from concerns about the practical utility of such features, privacy implications, or the potential for future costs, an issue previously highlighted when Samsung indicated (a past statement) that some Galaxy AI features from the S24 (a device launched in early 2024, a past event) might transition to a paid model after 2025.

Following a period of earlier setbacks, Google's AI vision has now come into sharp focus, signaling a cohesive strategy centered on its Gemini models to challenge competitors like OpenAI, CNET said in a report. This refined approach, which became evident following announcements at its Google I/O event in May 2024, involves deep integration of AI across Google's product suite, promising enhanced user experiences and intensifying competition within the AI industry.

AMD has purchased Enosemi, a startup that designs custom materials to support the development of silicon photonics products, in a strategic move to bolster its ambitions in the artificial intelligence sector, TechCrunch said in a report. This acquisition aims to enhance AMD's capabilities in developing high-speed, low-power optical interconnects, which are increasingly vital for improving data transfer speeds and efficiency within large-scale AI models and data centers, thereby strengthening its competitive position in the AI hardware market.

Ask what—if anything—is holding back the AI industry, and the answer you get depends a lot on who you’re talking to. I asked one of Bloomberg’s former chief data wranglers Carmen Li, and her answer was “price transparency.”

According to Li, the inability of most of the smaller AI companies to predict how much they will need to spend for the privilege of renting time on a GPU to train their models makes their businesses unpredictable and has made financing AI companies unnecessarily expensive. She founded the startup Silicon Data to create a solution: the first worldwide rental price index for a GPU.

That rental price index, called the SDH100RT, launched today. Every day, it will crunch 3.5 million data points from more than 30 sources around the world to deliver an average spot rental price for using an Nvidia H100 GPU for an hour. (“Spot price” is what a commodity to be delivered immediately sells for right now.)

“I really believe compute will be the biggest resource for humanity in the next few years,” says Li. “If my thesis is right, then it will need more sophisticated risk management.”

According to Li, such an index will lead to cheaper AI tools and more opportunities for a wider set of players to get involved in the AI industry. How do you get from an index to all that? Silicon Data’s origin story helps explain it.

US $1.04: Rental price advantage for Nvidia H100 GPUs on the East Coast of the United States versus those on the West Coast.

Until early last year, Li was in charge of global data integration at Bloomberg. In that position she met with several small companies that were trying to deliver AI-fueled data products, and many of them were struggling with the same problem. They could only offer their product at a fixed rate, but the cost of the GPU time they needed was unpredictable. Therefore, so were their profit margins.

With typical commodities like energy, companies can plan for these swings by knowing historical trends and hedging with financial products like futures contracts. But that didn’t exist for AI’s main commodity: time on a GPU. So Li set out to create the foundation for those products, and the result is the SDH100RT price index.

She chose to index the Nvidia H100, because it’s the most widely deployed GPU, and it’s used to train new AI models. However, a price index for Nvidia A100s, which tackle a lot of inference tasks, is in the works as well. And she’s developed a method that will determine when it makes sense to index prices for other AI chips, such as those from AMD and Nvidia’s Blackwell series.

Carmen Li founded Silicon Data after a stint at Bloomberg.Silicon Data

Armed with the data, startups and others building new AI products will be able to understand their potential costs better, so they can set their services at a profitable price. And those building new AI infrastructure will be able to set a benchmark for their own revenue. But just as important, in Li’s opinion, is that new sources of capital can get involved in the AI space.

Banks, for example, are a relatively inexpensive supplier of capital, notes Li. But because they have strict risk controls and there hasn’t been enough GPU price data, they haven’t been in a position to fund AI projects. Li hopes that the SDH100RT will let banks lend to a wider set of players in the AI industry and allow them to come up with financial products that reduce the risk for those already in it.

Insights and Oddities from the Data

Although it launched today, Silicon Data has been tracking GPU rental prices for months. As you might expect, having a window into the price of AI training has unveiled some interesting insights. What follows are a few things Li has discovered. (She’s been publishing these analyses on the regular since last September.)

East Coast rules! West Coast drools: H100 rental pricing is very stable in the United States, but there’s a persistent East Coast advantage. In March you could get an hour of work from an H100 on the East Coast for US $5.76. But that same hour would cost you $6.80 on the West Coast.

Hyperscaler chips help: Amazon Web Services’ foray into designing its own chips and servers has lowered prices for the cloud giant’s customers. According to Silicon Data, at about $4.80 per hour, the average unit price per GPU for AWS’s Trainium2 is less than half the price for using an Nvidia H100. Its first-generation chips Inferentia and Trainium both come in at less than $1.50 per hour, which is less than half the price of today’s inference workhorse, the Nvidia A100. However, H100s are thought to be the only option for cutting-edge model training, so their performance might justify the extra scratch.

DeepSeek’s modest effect: January’s DeepSeek shock did little to the spot rental price. You may recall that the performance and reported low-cost training of Hangzhou-based DeepSeek’s LLMs took many by surprise and sent AI-related stocks into a patch of pearl clutching. “When DeepSeek came out, the [stock] market went nuts,” says Li. “But the spot price didn’t change much.” On DeepSeek’s debut, the H100 price went up mildly to $2.50 per hour, but that was still in the $2.40 per hour to $2.60 per hour range from the months before. It then slid to $2.30 per hour for much of February before it started climbing again.

Intel is more posh than AMD: GPUs are always under the control of CPUs, usually in a 4:1 ratio. And the market for that CPU spot is contested between Intel and AMD. (Nvidia also makes its own CPU, called Grace.) But it seems customers are willing to pay a bit of a premium for Intel-powered systems. For Nvidia A100 systems, those with Intel CPUs fetched about a 40 percent higher price than those with AMD processors. For the H100, the effect depended on the interconnect technology involved. If a computer used SXM or PCIe as its links, Intel fetched a higher price. But for those using Nvidia’s NVLink interconnect scheme, AMD got the premium.

The Commoditization of AI

Can you really boil the price of AI down to a single number? After all, there are so many factors involved in a computer’s performance and its utility to a particular customer. For example, a customer might be training with data that cannot, for legal reasons, cross international borders. So why should they care about the price in another country? And, as anyone who has examined machine learning’s leading benchmark results, MLPerf, can see, the performance of the same Nvidia GPU can vary widely depending on the system it’s in and the software it’s running.

According to Li, the commodity view can work. Silicon Data’s index normalizes all these differences and gives different weights to things like how much a data center participates in the rental market, its location, its data sources, and many, many other things.

Perhaps the biggest endorsement of the idea of AI as a commodity is from Nvidia CEO Jensen Huang himself. At the company’s big developer event, GTC, he pushed for thinking of data centers as “AI factories” whose output would be measured in how many tokens, the smallest unit of information an LLM uses, they can produce per second.

Uplinq, a company specializing in AI-powered bookkeeping and tax solutions for small and medium-sized businesses (SMBs), announced on May 29, 2025, that it has secured $10 million in a Series A funding round, as per a report in AP. This new capital is intended to accelerate Uplinq's expansion, further develop its artificial intelligence technologies, and scale its efforts to acquire new customers.

Grammarly, known for its AI-driven writing assistance tool, has recently obtained a significant $1 billion commitment in nondilutive funding from General Catalyst, TechCrunch said in a report. This type of financing provides Grammarly with substantial capital without requiring the company to issue new shares or dilute existing shareholder equity. The funds are expected to be utilized for accelerating research and development, enhancing product features, particularly advanced AI capabilities, exploring potential market expansion, and making strategic investments. This development is also anticipated to intensify competition within the AI writing and productivity market, as Grammarly is now better positioned to innovate and potentially increase its market share.

Nvidia, the AI and graphics chip company, reported its first-quarter revenue increased by 69% from a year ago, beating financial analysts' estimates, VentureBeat says in a report. This substantial growth is primarily attributed to the soaring demand for Nvidia's data center GPUs, which are critical for AI processing workloads and are enabling the company to drive societal changes with artificial intelligence.

Nvidia is anticipating a multi-billion dollar revenue loss after taking a $4.5 billion charge in the first quarter of 2025, a direct consequence of United States licensing requirements that affect its ability to sell the H20 artificial intelligence chip to Chinese firms, TechCrunch said in a report. This significant charge is understood to be primarily an inventory write-down or a provision for potential losses concerning the H20 chips, which were specifically developed by Nvidia to comply with US export controls (first implemented in October 2022 and updated in October 2023) for the Chinese market. The financial hit underscores the challenges Nvidia faces in navigating these regulations and the market reception for its compliant chips in China, amidst a backdrop of reported weak demand for such modified hardware and an acceleration of domestic Chinese AI chip development.

Mistral recently unveiled Codestral Embed, a new embedding model specifically for code, which aims to accelerate Retrieval Augmented Generation (RAG) use cases and enable the identification of duplicate code segments using natural language, as per a report in VentureBeat. The company claims this model surpasses competitors like OpenAI and Cohere in performance on real-world retrieval tasks. This development highlights the growing trend of specialized AI models and intensifies competition in the market for AI tools tailored to coding, potentially offering developers more effective solutions for managing and understanding large codebases.

TechCrunch Sessions: AI is spotlighting the future of artificial intelligence and the startups shaping it, bringing together founders, investors, and technologists to explore how AI is transforming industries from the pitch deck to full-scale deployment, TechCrunch said in a report. The event, described as high-energy, also aims to highlight the contributions of its partners in this evolving landscape. It's generally understood that while AI is profoundly transformative, its widespread integration is typically a complex, phased process, and innovation in the field involves contributions from startups as well as large tech companies and academic institutions.

Thailand's Ministry of Digital Economy and Society (MDES) and Huawei Cloud jointly launched an initiative at the Huawei Cloud Summit Thailand 2025 on May 30, 2025, to accelerate Thailand's agenda to become the AI hub in ASEAN, AP said in a report. During the summit, MDES also shared progress made over the past year on its "Cloud First" strategy, which aims to foster an inclusive, sustainable, and opportunity-filled Thailand by encouraging cloud adoption in both public and private sectors. This new AI initiative aligns with Thailand's ongoing National AI Strategy and Action Plan (2022-2027).

Snowflake is tackling key enterprise AI deployment hurdles by releasing new open-source text-to-SQL solutions and advancements for its Arctic inference models, VentureBeat said in a report. These recent developments are designed to simplify how businesses convert natural language questions into database queries and to enhance the performance and efficiency of running large language models for inference tasks within enterprise systems. The company's strategy aims to make AI more accessible and practical for its customers by addressing these critical operational pain points.

The recent takedown of the DanaBot malware, responsible for over $50 million in damages, was significantly accelerated by the use of agentic AI, which reduced complex security operations center analysis from months to weeks, VentureBeat said in a report. This application of AI systems, designed to autonomously plan and execute multi-step tasks, proved decisive in dismantling the Russian malware platform. The success highlights the potential for agentic AI to revolutionize cybersecurity by speeding up investigations and enabling more effective responses to sophisticated threats, likely spurring further development of autonomous capabilities in security tools.

A-share tech-focused Exchange-Traded Funds (ETFs) in China have continued to attract significant investor interest in 2025, with the top five industry/thematic ETFs by net inflows as of May 21 all being technology-related, collectively drawing in US$7.87 billion, AP said in a report. Among these, the E Fund CSI Artificial Intelligence ETF (159819) saw a notable net inflow of US$1.17 billion. This trend is accompanied by leading asset managers in China actively positioning themselves in such tech-focused ETFs, reflecting increased investor confidence in the mainland Chinese stock market and the growth potential of sectors like AI, which are also a strategic focus for the Chinese government.

Delaware Attorney General Kathy Jennings intends to hire an investment bank to advise her office as it scrutinizes OpenAI's restructuring strategies designed to draw more investment capital, Bloomberg said in a report. This move underscores Delaware's significant regulatory role, as OpenAI is incorporated in the state, granting the Attorney General authority over its corporate governance. The investment bank will offer specialized financial expertise to assess the complexities and potential consequences of OpenAI's plans, which likely involve changes to its corporate and equity structure to facilitate new funding and employee liquidity.

Chinese startup DeepSeek announced and subsequently released an updated version of its R1 reasoning AI model on the developer platform Hugging Face, as per a report in TechCrunch. The company described this update, communicated via a WeChat message on Wednesday morning, May 28, 2025, as "minor." Importantly, the updated R1 model is available under a permissive MIT license, which explicitly permits commercial use.

The New York Times and Amazon have announced a significant multi-year licensing agreement, reportedly valued at over $100 million, allowing Amazon to use The Times' current and archived editorial content for training its artificial intelligence models and for display on Amazon platforms, according to a report in New York Times. This deal, which provides The Times with a new revenue stream and Amazon with high-quality training data, occurs while The Times' (past event) December 2023 copyright infringement lawsuit against OpenAI and Microsoft remains ongoing and separate. The agreement underscores a growing trend of news publishers engaging in licensing deals with technology companies developing AI.

Nvidia Chief Executive Officer Jensen Huang stated that Chinese AI rivals are increasingly powerful and are effectively occupying the market space in China vacated by US companies due to American export controls on advanced AI technology, as per a report in Bloomberg. This situation, where US firms are restricted from selling their most advanced AI chips (a measure significantly updated in October 2023), has allowed domestic Chinese firms to strengthen. Huang's recent assessment indicates that these Chinese entities are now significant challengers, with their technology becoming more potent in the AI sector.

North Carolina recently became the fourth U.S. state to adopt an Artificial Intelligence (AI) guidebook for its K-12 students, a development intended to be helpful for both teachers and students in educational settings, as per a report in CNN Technology. Coy Wire, in his current role with CNN 10, is reported to be explaining how this guidebook can be beneficial. This adoption is part of a larger trend across U.S. states establishing official AI education guidelines and reflects an increased national focus on equipping students and educators with AI literacy. The guidebook is anticipated to offer students structured learning on AI principles and ethical usage, while providing teachers with resources for integrating AI into their curriculum and managing its classroom application.

Horizon3.ai, a cybersecurity startup specializing in autonomous penetration testing tools, is aiming to raise $100 million in a new funding round and has already secured at least $73 million, TechCrunch said in a report. The company disclosed this progress in an SEC filing this week, and according to two people familiar with the deal, New Enterprise Associates (NEA) led the investment round.

ValueLabs, a global leader in Agentic AI Services, announced on May 29, 2025, a significant strategic change, transitioning entirely from traditional Time & Materials (T&M) contracts for new clients across its major global markets, as per a report in AP. This move underscores the company's dedication to delivering measurable outcomes and leveraging innovation, particularly through its Enterprise OS platform, AiDE™, which is designed to power these AI-driven delivery models. This change applies specifically to new client engagements, with existing contracts likely continuing under their current terms. This development occurs within a broader IT industry context, noted since late 2024, of a growing trend towards outcome-based pricing, increased AI adoption in service delivery, and platform-led service models. For ValueLabs, this positions them as a potential market leader but also introduces higher financial risk if outcomes are not met efficiently, while new clients gain predictable costs tied to business value.

Vietnam's aspirations to become a regional AI leader heavily rely on the success of FPT Corp., its largest tech firm valued at $6.8 billion, as it undertakes a challenging global expansion journey, as per a report in Bloomberg. The company is currently confronting stiff competition from established international rivals and navigating geopolitical risks, such as potential US tariffs, which could impact its access to key markets and the cost of essential components. FPT's growth, which was supported by strategic acquisitions in the 2023-2024 period (a past activity), is central to Vietnam's national AI strategy (a past initiative approved in 2021 aiming for leadership by 2030), though the broader tech ecosystem also contributes. FPT's current performance is critical not only for national goals and shareholder value but also signals Vietnam's increasing presence in the global IT and AI services market.

Artificial intelligence tools, including ChatGPT, pose a risk of exposing users' personal data, which scammers can then exploit, according to a report in Fox News. Tech expert Kurt “CyberGuy" Knutsson, in his current role, advises on these emerging privacy risks and how individuals can protect themselves. The danger primarily arises not from the AI model itself directly providing data to scammers, but from users inputting sensitive or personal information into AI queries, combined with potential security vulnerabilities or data handling issues on the platform hosting the AI, which could lead to data breaches or accidental exposure.

Google is entering the "vibe coding" field with Stitch, a new tool designed to create user interfaces from a single prompt, as per a report in VentureBeat. This tool, a follow-up to Google's earlier project Jules, aims to generate front-end UI code based on natural language descriptions, reflecting a growing trend where AI assists in abstracting code generation. The development of Stitch places Google in competition with existing generative UI platforms and aligns with broader advancements in AI for code generation and its integration into design tools, trends that have been evolving since 2023. "Vibe coding" itself signifies a shift towards generating UIs from high-level, potentially subjective prompts rather than detailed technical instructions.

Microsoft co-founder and philanthropist Bill Gates recently elaborated on the transformative potential of artificial intelligence, particularly its capacity to improve millions of lives through advancements in healthcare and education, as per a report in CNN. He shared these insights during a conversation with Fareed Zakaria, emphasizing the positive impact AI could have in these critical sectors.

Hume AI has introduced its EVI 3 model, which allows for the customization of AI-generated voices, focusing on humanlike emotional expression and intonation, ZDNet said in a report. The company believes that AI models capable of speaking and emoting convincingly will define the future of artificial intelligence interaction, offering capabilities to adjust vocal parameters such as emotional tone and style. This development is anticipated to provide content creators with more versatile audio tools and could lead to more empathetic AI assistants in customer service.

A government report, identified in context as likely the Department of Homeland Security's AI Safety Advisory Board report from May 2024, warns that advanced Artificial Intelligence systems could pose an 'extinction-level threat' to humans and that the US must intervene, as per a report in CNN Technology. Ret. Admiral James Stavridis, who currently serves as an Operating Executive at The Carlyle Group, commented on this warning, stating, 'I think we should be mindful of it,' but also added historical context: 'there have been big inventions in the past - the printing press, electricity, the internet - all of these have been a decried for the possibility of nefarious activity.'

MIT Technology Review is continuing its in-depth examination of the substantial energy consumption expected to fuel the growth of artificial intelligence, as per a report in MIT Technology Review. This follows the publication just over a week ago of their "Power Hungry" package, with a recent newsletter breaking down a key analysis from that package concerning AI's power needs, conducted by one of its journalists and colleague James O’Donnell.