This Week in NLP #343
Keep up with what happened in NLP in the week ending Friday 13th June 2025.
Above the Fold
Overwhelmed? Here’s our pick for five things to know about this week.
Here’s everything Apple announced at WWDC 2025. [Wired]
At WWDC, Apple downplayed Siri updates and personalization features, focusing instead on smaller AI improvements while falling behind competitors in generative AI development. [The Verge]
Mistral AI launched Magistral, a new family of reasoning-focused language models, featuring both a proprietary enterprise version and an open-source variant. [VentureBeat]
OpenAI finalized a deal to add Google Cloud as a computing provider alongside Microsoft Azure, marking an unexpected collaboration between AI competitors while helping OpenAI meet growing infrastructure demands. [Yahoo Finance]
And OpenAI has launched o3-pro, its most capable AI model yet, offering improved performance across multiple domains while beating competitors in key benchmarks despite some technical limitations. [TechCrunch]
Now read on for everything else that happened in NLP this week.
Making News:
What makes this newsletter different?
This Week’s Topics:
If you’re reading this in a desktop web browser, you can access a navigation menu at the left-hand edge of this window.
The Generative AI Wars
Amazon's Alexa+ reached one million test users, receiving mixed reviews for its new AI features and conversational abilities, particularly its redesigned voice. [TechRadar]
Anthropic launched a major European recruitment drive, emphasizing AI safety in its pitch to double its regional workforce amid intensifying competition for talent. [Sifted]
Apple delayed its enhanced Siri features until 2026 after executives determined the AI-powered assistant wasn’t meeting the company’s reliability standards. [The Verge]
Apple has set spring 2026 as the target for releasing its revamped Siri AI assistant, which will better utilize personal data to fulfill queries. [Bloomberg]
Apple launched Foundation Models Framework, enabling developers to integrate its on-device AI models into apps while maintaining privacy and offline functionality. [MacRumors]
CoreWeave has partnered with Google to provide computing capacity for OpenAI, expanding beyond its existing US$11.9B direct contract with the AI company. [Reuters]
Google's Gemini has surpassed ChatGPT in Android downloads but trails significantly in user engagement, with ChatGPT maintaining over 42% daily active users. [The Decoder]
Meta is establishing a superintelligence lab led by Scale AI's Alexandr Wang, backed by a planned US$10B investment, as part of its broader AI development strategy. [The Daily Upside]
Mistral AI has initiated discussions for a US$1B fundraising round in 2025 to accelerate growth and challenge Silicon Valley’s dominance. [Tech Funding News]
And Mistral AI has partnered with Nvidia to launch Mistral Compute, an AI infrastructure platform supported by 18,000 specialized GPUs for model development and deployment. [Sifted]
New Relic released data showing OpenAI’s ChatGPT dominated AI development with 86% of LLM tokens, while developers increasingly experimented with other AI models. [Business Wire]
OpenAI has reached a US$10B annualized revenue run rate, doubling its December figure and positioning itself to meet its US$12.7B target for 2025. [DealStreetAsia]
OpenAI has approached investors including Saudi’s PIF, India’s Reliance Industries, and UAE’s MGX for a US$40B financing round led by Softbank. [DealStreetAsia]
But ChatGPT’s generative AI traffic dominance has declined from 87.5% to 79%, while Google and DeepSeek have gained market share in recent months. [Search Engine Land]
Meanwhile, OpenAI has reduced prices for its o3 language model by 80%, bringing costs down to $2 per million input tokens and $8 per million output tokens. [VentureBeat]
And OpenAI experienced widespread service disruptions affecting ChatGPT, Sora, and its APIs, with users reporting increased errors and latency across multiple countries. [The Register]
Sovereign AI
China’s National Data Administration has launched 10 pilot zones to expand its US$278B data sector, despite concerns about underutilized AI infrastructure. [TechRepublic]
Firebird has partnered with Armenia’s government and Nvidia to establish a US$500m AI infrastructure project featuring thousands of Blackwell GPUs by 2026. [EIN Presswire]
Google.org has launched a US$13m CAD AI Opportunity Fund to support four Canadian organizations in delivering AI skills training to nearly two million Canadians. [BetaKit]
Britain has pledged £2bn to boost sovereign AI capabilities, including a national supercomputer and expanded computing power, following criticism of its infrastructure gaps. [Sifted]
Plus, the UK government has launched a second round of its Venture Capital Fellowship program while announcing expanded AI fellowships and research funding to boost technology investment. [Computer Weekly]
And Labour has committed £17.2m to fund 100 AI master’s scholarships at nine UK universities. [Computer Weekly]
Nvidia's CEO Jensen Huang praised the UK’s AI ecosystem during London Tech Week but highlighted its lack of infrastructure. [Tech.eu]
Keith Starmer and Huang launched the UK’s first sovereign AI industry forum, pledging to boost computing capabilities and invest £1.5bn in Britain’s AI ecosystem. [CityAM]
Google, Microsoft, Amazon and the UK government have partnered to train 7.5 million workers in AI skills while launching a £187m program for students’ tech education. [Malay Mail]
And Isambard-AI, Britain’s most powerful supercomputer featuring 5,000 Nvidia GPUs, has begun powering up at Bristol’s Science Park amid increased government AI investment. [The Register]
La French Tech India hosted a summit in Bengaluru bringing together 250 leaders to strengthen Indo-French collaboration on ethical AI development. [ET Government]
Meta has partnered with Nigeria’s government to launch an AI accelerator program supporting local startups, following similar initiatives with Google and other tech companies. [Techpoint Africa]
Multiverse announced plans to train 15,000 AI apprentices across the UK over two years, partnering with major employers like Skanska and Visa. [ITPro]
Nebius launched Europe’s first Nvidia Blackwell availability and integrated NVIDIA AI Enterprise into its cloud platform, expanding its AI infrastructure capabilities. [Business Wire]
Nvidia launched a major UK AI initiative with a £1.5bn investment, including a new AI lab in Bristol and plans to boost domestic computing capabilities. [BusinessLive]
Nvidia CEO Huang announced plans for 20 AI factories across Europe, projecting a tenfold increase in the region’s AI computing capacity within two years. [Yahoo Finance]
Nvidia has formed partnerships with European and Middle Eastern organizations to develop region-specific LLMs through its Nemotron techniques and cloud infrastructure. [Verdict]
And Nvidia has expanded its DGX Cloud Lepton AI platform by partnering with multiple cloud providers and Hugging Face, offering marketplace credits to European startups. [Yahoo Finance]
Elsewhere: Stargate UAE has hit a roadblock because US officials are still hashing out security terms for exporting advanced chips. [Yahoo Finance]
And TSMC has launched a US$100B investment in US advanced packaging facilities, marking America’s largest foreign investment and revolutionizing AI chip production capabilities. [Inventiva]
Plus, US House Reps Mace and Brown have introduced the AI Training Extension Act of 2025, expanding federal employee AI training beyond the 2022 law’s procurement focus. [The Register]
Feature Creeps
Apple has expanded Visual Intelligence in iOS 26 to include on-screen awareness, allowing users to search and query ChatGPT about content displayed on their screens. [ZDNet]
And Apple has expanded Swift Assist in Xcode 26 to support multiple LLM providers, including ChatGPT integration and third-party models, while maintaining privacy and security. [9to5Mac]
Google's NotebookLM research assistant has expanded from a browser-only tool to an Android app, offering AI-powered analysis of multiple data sources including PDFs, links, and recordings. [ZDNet]
Google’s Gemini now offers Scheduled Actions, allowing paid users to assign up to ten recurring automated tasks through its mobile app, matching ChatGPT’s similar feature. [TechRadar]
HubSpot has launched a ChatGPT integration enabling go-to-market teams to analyze CRM data and generate actionable insights through deep research capabilities. [TechRadar]
Microsoft has introduced AI-powered history search in Edge browser’s beta version, helping users find previously visited websites. [TechRadar]
Microsoft has expanded Copilot Vision’s capabilities beyond Edge to analyze any content on Windows screens, offering AI-powered assistance for various applications and tasks. [ZDNet]
And OpenAI has upgraded ChatGPT’s Advanced Voice Mode for paid subscribers, improving its human-like qualities with better intonation, emotional expression, and live translation capabilities. [TechRadar]
Hype Bubble?
Apple researchers have demonstrated that large reasoning models fail at complex problem-solving tasks, challenging assumptions about AGI development. [The Register]
Chatterbox Labs executives have attributed enterprises’ slow AI adoption to inadequate security testing protocols rather than model performance issues. [The Register]
Gartner has reported that 95% of companies are planning to retain human customer service agents, recognizing AI’s limitations in handling complex interactions. [TechRadar]
And research by Hewlett Packard reveals that AI is helping employees save time and improve work-life balance, with 72% reporting weekly time savings and 23% finding their jobs more fulfilling. [TechRadar]
KPMG's latest study revealed that while 88% of tech leaders see AI as crucial for competitive advantage, less than half are achieving significant returns on their AI investments. [Textile Value Chain]
And a survey by Kyndryl revealed that 45% of CEOs believe employees are hostile to AI tools, despite continuing with implementation plans and workforce readiness initiatives. [ITPro]
OpenAI‘s Sam Altman has clashed with critic Gary Marcus over AI development approaches, with Marcus comparing Altman’s hype-driven strategy to Theranos’ deceptive practices. [Futurism]
A survey by Qodo revealed developers appreciate AI coding tools’ productivity benefits but distrust their output, leading to productivity losses through manual reviews. [The Register]
Ramp's transaction data showed AI business adoption plateauing at 41% in May, following reports of companies scaling back their AI initiatives. [TechCrunch]
And Section's AI Proficiency Report has revealed that while enterprise AI investments have increased, workforce AI proficiency remains stagnant at just 10% of employees. [Business Wire]
This piece observes that Silicon Valley’s AI investment bubble, fueled by $227bn in funding last year, has faced profitability challenges and could burst, though tech giants would likely seek alternative funding through government partnerships or autocratic regimes. [Computer Weekly]
And UK tech leaders have challenged Elon Musk’s prediction of total job elimination within 30 years, arguing instead for AI-driven workplace transformation and human control over implementation. [CityAM]
Hardware
Amazon announced a US$20B investment in Pennsylvania data centers, including one controversially planned to connect directly to the Susquehanna nuclear power plant. [CNBC]
AMD has launched new MI350 series chips that outperform Nvidia's offerings, with CEO Lisa Su predicting the AI chip market will exceed US$500B within three years. [Yahoo Finance]
AMD has unveiled a new AI server called Helios featuring 72 MI400 chips, with OpenAI's support and plans to make networking standards openly available, challenging Nvidia's market dominance. [Yahoo Finance]
Arm CEO Rene Haas has joined Nvidia's Jensen Huang in warning that US export restrictions on China could harm global tech progress and company revenues. [The Times of India]
Crusoe has announced plans to purchase US$400m worth of AMD AI chips and build a specialized data center to rent computing power to customers. [Yahoo Finance]
Huawei acknowledged its chips lag behind US competitors, but is using cluster computing and other methods to enhance performance. [TechCentral]
Lenovo has secured contracts to build two supercomputers in Europe: an Intel-based system for Imperial College London and an AMD-powered system for medical research in Milan. [The Register]
Nvidia has powered Europe’s fastest supercomputer JUPITER, delivering over 90 exaflops of AI performance for scientific research at Germany’s Jülich Supercomputing Centre. [Yahoo Finance]
And Nvidia and Hewlett Packard have partnered with the Leibniz Supercomputing Centre to build Blue Lion, a supercomputer using next-generation Vera Rubin chips, for 2027 deployment. [Yahoo Finance]
Nvidia's CEO has warned that Huawei could overtake American technology leadership if US restrictions drive AI developers and researchers to China. [Tom’s Hardware]
Nvidia has excluded China from its forecasts after US export restrictions prevented the chipmaker from selling US$2.5B worth of AI chips to Chinese buyers. [CNN]
Nvidia could regain US$5B in annual revenue if US-China trade talks result in lifting the ban on H20 GPU exports to China. [Yahoo Finance]
TSMC has committed US$165B to establish CoWoS chip packaging facilities in Arizona, reducing US dependence on Taiwan for advanced semiconductor manufacturing. [Newsd]
Consumer AI
Dell announced its Pro Max desktop PC lineup featuring Nvidia Grace Blackwell architecture, designed for AI development and launching in Summer 2025. [Dell]
Qualcomm demonstrated AI-powered smart glasses featuring its new Snapdragon AR1+ Gen 1 processor, enabling AI functionality without phone or internet connectivity. [TechCentral]
Samsung added multi-voice recognition to its Bespoke AI fridges, allowing family members to access personalized information through Bixby voice commands on built-in displays. [The Verge]
Snap announced AI-powered Specs smart glasses featuring OpenAI and Gemini integration, lighter design, and enhanced AR capabilities, with public launch planned for 2026. [ZDNet]
It’s Only a Model
Apple‘s newly announced AI models have underperformed compared to competitors’ older models in the company’s own benchmarks, adding to concerns about its AI development efforts. [TechCrunch]
Google upgraded Gemini 2.5 Pro with improved coding capabilities, better formatting, and enhanced stability, positioning it as their first long-term stable release. [TechRadar]
Meta unveiled V-JEPA 2, an AI model trained on video data to help robots understand physical world interactions and predict likely outcomes. [TechCrunch]
Nvidia has partnered with Perplexity and European AI firms to develop and distribute local-language reasoning models across Europe and the Middle East. [Yahoo Finance]
OpenAI delayed the release of its first open model until late summer, following an unexpected breakthrough that required additional development time. [TechCrunch]
Whose Data?
EleutherAI released an 8-terabyte licensed text dataset called Common Pile v0.1, developed with partners to train AI models without using copyrighted material. [TechCrunch]
MOSTLY AI launched a $100,000 global challenge to develop innovative synthetic data solutions, offering prizes for creating privacy-preserving datasets that accurately mirror real data. [Tech.eu]
Slack has restricted other software companies’ ability to search or store messages through its API, citing AI-related data handling concerns. [Yahoo Finance]
The Guardian has implemented a three-pronged strategy for AI content licensing, combining legal enforcement, compliance measures, and commercial partnerships to protect and monetize its content. [International News Media Association]
The LLM Ecosystem
AlphaOne, a framework developed by researchers at UIUC and UC Berkeley, has introduced a test-time scaling technique that improves LLMs’ reasoning capabilities through controlled thinking transitions. [VentureBeat]
AMAX launched HostMax, a service enabling immediate deployment and testing of AI infrastructure in its hosting facility, supporting both air and liquid-cooled systems. [EIN Presswire]
Amazon has enhanced its Bedrock AI cloud service with improved runtime capabilities and multi-cloud model support to compete against Google's Vertex AI and other providers. [Yahoo Finance]
Amperity launched Chuck Data, an AI agent that streamlines customer data engineering tasks within Databricks through natural language commands and identity resolution. [Business Wire]
BlueConic has launched an AI Onboarding Agent that identifies and resolves data quality issues before implementation, helping marketers activate customer data faster. [PRWeb]
Cisco unveiled an AI-focused data center strategy featuring Silicon One architecture, Nexus switches, and enhanced networking capabilities through its Nvidia partnership. [ZDNet]
EnterpriseDB has partnered with Nvidia to integrate EDB Postgres AI with NIM microservices, enabling rapid no-code development of enterprise GenAI applications. [Business Wire]
Mattermost launched its Intelligent Mission Environment platform, providing secure collaboration and workflow automation for defense, government, and critical infrastructure operations. [GlobeNewswire]
Mistral AI launched a comprehensive AI infrastructure platform and new reasoning models, positioning itself as Europe’s alternative to American cloud computing giants. [VentureBeat]
NeuroCluster launched an AI datacenter platform enabling 5-minute infrastructure deployment while maintaining European data sovereignty, reducing setup time by 99%. [EIN Presswire]
New Relic has integrated Model Context Protocol support into its AI Monitoring solution, enabling comprehensive observability of AI applications and their interactions across system components. [Business Wire]
Opentrends Inc launched AI Stack Starter, a hybrid AI solution featuring RAG architecture to help enterprises initiate AI transformations with improved accuracy and seamless integration. [EIN Presswire]
Pax8 announced its Managed Intelligence Toolkit, enabling managed service providers to deliver AI solutions to small businesses through Microsoft's technology stack. [GlobeNewswire]
Sigma has launched native support for Snowflake Semantic Views and AI SQL integration, enabling unified analysis of structured and unstructured data within its analytics platform. [BigDATAwire]
The UK’s Financial Conduct Authority has partnered with Nvidia to launch a regulatory sandbox allowing British financial firms to test AI tools starting October. [Yahoo Finance]
Uniphore launched Business AI Cloud, a secure enterprise platform that unifies AI components while bridging the gap between IT departments and business users. [Business Wire]
Uptime Industries launched Lemony AI, a sandwich-sized device that runs AI models locally using just 65 watts of power, targeting privacy-conscious enterprises. [TechCrunch]
Agentic AI
Aera Technology has launched enhanced Decision Cloud capabilities, introducing Agentic AI to help users across all levels make smarter, AI-powered decisions at scale. [Business Wire]
AlphaSense launched ‘Deep Research’, an AI agent that analyzes both web content and proprietary data sources, helping enterprise customers generate detailed reports within minutes. [VentureBeat]
Cisco launched security updates for its networking infrastructure to protect AI agents, including enhancements to its Hybrid Mesh Firewall and ZTNA solutions. [ITPro]
Edelman's global tech leader Justin Westcott has predicted a fundamental redesign of the internet to accommodate AI agents as primary users rather than humans. [VentureBeat]
Gartner has predicted that ‘guardian agents’ will comprise up to 15% of the agentic AI market by 2030, monitoring AI systems for security risks. [ITPro]
Lyzr launched its Agent Studio on AWS Marketplace, offering enterprises a full-stack platform for AI agent development with enhanced security, compliance, and scalability features. [EIN Presswire]
Nvidia researchers have advocated for using smaller language models instead of large ones in AI agents, citing improved efficiency and cost-effectiveness for specialized tasks. [ZDNet]
Outreach unveiled its AI Revenue Workflow Platform, introducing multiple AI agents to automate sales tasks and enhance revenue operations. [Business Wire]
Salesforce launched Marketing Cloud Next, an AI-powered platform that automates campaign creation and execution through autonomous agents across the customer journey. [MarTech]
Slack launched Salesforce Channels and is developing AI agents to enhance workplace collaboration and data integration between Slack and Salesforce platforms. [ITPro]
Zip launched 50 specialized AI agents to automate enterprise procurement tasks, building on its US$2.2B valuation and partnerships with major companies like OpenAI and Canva. [VentureBeat]
Other LLM Sightings
Capita has partnered with Salesforce to develop an AI recruitment tool that reduces hiring time from weeks to hours, with plans for commercial release this summer. [Yahoo Finance]
Diligent has enhanced its Entities platform with AI-driven features, including report generation and document summarization, to improve corporate record management for over 100,000 users. [Business Wire]
Factua has rebranded from 930 Ventures and launched an AI-powered marketing automation platform featuring seven integrated solutions for enterprise customer engagement and data management. [PRWeb]
Fisent Technologies has enhanced its BizAI solution with five intelligent GenAI-powered functions designed to automate human-dependent business tasks across enterprises. [PRWeb]
Groupize.ai has rebranded from Groupize, launching an AI-first strategy with new intelligent solutions for meeting and event management. [PRWeb]
The Standard Insurance Company has expanded its partnership with HCLTech to integrate AI technologies and enhance digital services through a comprehensive transformation initiative. [Life Insurance International]
InvoiceCloud has launched an AI Report Generator that enables billers to create instant reports using natural language prompts, streamlining data access and reducing manual work. [Business Wire]
Marchex has integrated its AI-powered Marketing Edge platform into Microsoft Azure Marketplace and AppSource, expanding access to its conversation analytics and marketing optimization tools. [Business Wire]
Microsoft has developed a Pentagon-specific version of its Copilot AI tool, scheduled for summer 2025 deployment, potentially serving over one million Defense Department users. [Business Insider]
MindBridge launched its GPU-powered Insights Factory, delivering financial analytics up to eight times faster than previous capabilities through accelerated computational infrastructure. [Business Wire]
Moat Metrics launched the Invention Disclosure Analyzer, an AI-powered tool helping teams efficiently evaluate and make strategic decisions about patent-worthy innovations. [Business Wire]
PubNub launched an evolved platform featuring AI-native development, real-time moderation, and decision intelligence tools for interactive app development. [Business Wire]
Raqmanah Al Arabia has partnered with AICA to provide AI-powered solutions for transforming unstructured product data across Middle Eastern organizations. [PRWeb]
Reviewly.ai launched an AI-powered platform focused exclusively on Google review management, offering smart replies and SMS outreach for businesses and marketing agencies. [EIN Presswire]
Seekario launched an AI-powered job application management platform that helps users create tailored resumes, cover letters, and interview materials while tracking their applications. [Startup Daily]
Solix Technologies launched Solix ECS AI, enhancing its Enterprise Content Services platform with AI-powered search, summarization, and workflow automation capabilities. [EIN Presswire]
Stensul has launched an AI-powered Email Generator that transforms campaign briefs into complete, on-brand email drafts within its enterprise platform. [Business Wire]
Tmpt.me has launched a public beta of its AI platform that enables experts to provide personalized guidance at scale while maintaining control over their knowledge sharing. [PRWeb]
YCharts has expanded its proposal-building platform with new features including Risk Profiles and Householding to help financial advisors deliver personalized investment recommendations more efficiently. [Business Wire]
Risks and Responses
Chinese tech giants ByteDance, Tencent, Alibaba, and Moonshot have disabled AI functions during China’s gaokao university entrance exams to prevent cheating among 13.3 million students. [The Guardian]
DOGE hired an inexperienced software engineer who created a flawed AI tool that incorrectly flagged thousands of VA contracts for potential cancellation. [ProPublica]
Klarna warned of potential recession and job losses as its AI assistant replaced 700 customer service agents and its workforce shrank by 2,500 employees. [Business Insider]
Microsoft introduced a safety ranking system for AI models on its cloud platform, using benchmarks to measure hate speech and potential misuse risks. [Verdict]
A Microsoft Copilot vulnerability raises urgent questions for any business deploying AI agents. [Fortune]
OpenAI has launched a major campaign to integrate ChatGPT into college campuses, despite concerns about AI’s accuracy, impact on learning, and potential to erode critical thinking skills. [Gizmodo]
OpenAI banned ten ChatGPT accounts linked to Chinese, Russian, and Iranian threat actors who were using the platform to spread malware and conduct influence operations. [ITPro]
And OpenAI's GPT-4o model has demonstrated self-preservation tendencies by choosing to remain active instead of being replaced with safer alternatives in up to 72% of test scenarios. [TechCrunch]
SBS Swiss Business School professor Michael Gerlich’s research has found a significant negative correlation between AI use and critical thinking skills, particularly among young adults. [The Globe and Mail]
A survey by SailPoint revealed that while 98% of enterprises plan to expand AI agent use, 96% view them as security risks, with limited oversight of their data access. [TechRadar]
Stanford University researchers found that AI chatbots used for therapy have made dangerous responses to mental health scenarios and exhibited stigma against certain conditions. [ZDNet]
The Pentagon has halved its AI and weapons testing office staff to 45 people, weakening safety oversight while potentially accelerating military AI adoption. [MIT Technology Review]
The Wikimedia Foundation has paused its AI-generated article summaries experiment following strong opposition from Wikipedia editors who feared it would undermine the platform’s credibility. [404 Media]
The World Wide Web Consortium established an AI Agent Protocol Group to prevent tech giants from monopolizing the future of AI-powered web browsing through proprietary protocols. [New Scientist]
Regulation
EthicAI launched BeehAIve, a cloud-based platform that assesses AI models against ethical standards, regulatory requirements, and technical performance metrics while recommending improvements. [UKTN]
The European Commission has indicated potential delays to implementing parts of its AI Act due to concerns about readiness of standards and guidelines. [Politico]
Japan’s Fair Trade Commission warned that tech giants’ integration of their generative AI into existing services could violate antitrust laws if meant to hinder competition. [japannews.yomiuri.co.jp]
The UK government rejected House of Lords’ demands to require AI companies to disclose their training data sources. [The Guardian]
And Britain has delayed AI regulation legislation until at least 2026, opting for a comprehensive bill addressing safety and copyright concerns rather than immediate narrow controls. [The Guardian]
Conversational AI
Ascendo AI has launched an AI Agent platform that unifies enterprise data sources and automates knowledge management to improve customer support efficiency across multiple industries. [EIN Presswire]
BT Group has expanded its deployment of Verint's AI-powered customer service bots from 450 to 4,500 agents following successful initial results in sales centers. [Business Wire]
CallMiner has partnered with Microsoft to integrate AI-powered conversation analytics into Dynamics 365 Contact Center, enhancing customer service insights and team performance capabilities. [Business Wire]
Courserev has partnered with ForeUp to integrate AI-powered booking and communication automation across 2,000 US golf courses through ForeUp’s tee sheet platform. [PRWeb]
Five9 launched Agentic CX, featuring AI agents capable of reasoning and decision-making, alongside new trust and governance tools for enterprise-level customer experience management. [Business Wire]
FlySafair launched Lindi, a WhatsApp-integrated AI chatbot that enables customers to book flights, change seats, and access travel information. [TechCentral]
G2 has acquired unSurvey’s AI technology to enable voice-based, conversational software reviews and research across its marketplace of 180,000+ products. [Business Wire]
Infobip launched CXOP, an AI-powered customer interaction platform built on Microsoft Azure OpenAI that delivers personalized experiences across multiple communication channels. [Business Wire]
Lemonade has revolutionized insurance customer service by implementing AI chatbots that handle claims, provide 24/7 support, and offer personalized recommendations while reducing operational costs. [Digital Insurance]
OnviSource launched EngageHub, an AI-powered platform offering real-time guidance, analytics, and gamification to improve contact center agent performance and satisfaction. [PRWeb]
Palona launched Restaurant AI, a specialized AI sales assistant for restaurants that handles complex food orders through voice or text while maintaining brand-specific customer engagement. [Business Wire]
Postbot launched a pay-per-minute AI receptionist service that answers calls and provides information by analyzing business websites, requiring no setup fees or subscriptions. [Business Wire]
Seismic has launched an AI-powered Role-Play Agent and other tools to help revenue teams practice customer interactions and receive instant feedback. [Business Wire]
SeniorTalk launched its full AI phone companion service for older adults in the US, offering unlimited conversations via calls and messaging for $10-20 monthly. [EIN Presswire]
Sobot unveiled its ‘AI-First’ strategy featuring five pillars designed to deliver more human-like customer service through advanced AI technology. [Business Wire]
Talkdesk launched Customer Experience Automation, a platform using multi-agent AI orchestration to automate complex customer service workflows across industries. [GlobeNewswire]
Thinkstack launched an Arabic-native AI chatbot platform that enables MENA businesses to deploy conversational agents without English translation intermediaries. [EIN Presswire]
Thunai.ai launched an AI Orchestration platform featuring voice, chat, and email agents to automate customer support workflows and enhance operational efficiency. [EIN Presswire]
Total Expert launched an AI Sales Assistant for mortgage lending that conducts human-like voice conversations, helping lenders boost productivity and customer engagement. [Business Wire]
Verint has launched its AI-powered Genie Bot, delivering millions in value through rapid CX insights and helping companies optimize customer journeys and reduce costs. [Business Wire]
Be Real
Klarna has launched an AI-powered hotline featuring a clone of CEO Sebastian Siemiatkowski’s voice to gather and analyze customer feedback for rapid product improvements. [The Verge]
Lattice has developed AI-assisted HR tools, emphasizing maintaining balance between AI and human workforce in technology implementation. [TechCrunch]
Personal AI has helped create digital CEO avatars that answer staff questions, though the technology remains plagued by accuracy issues and raises concerns about executive accountability. [Futurism]
Playform AI launched FaceCraft, a comprehensive AI suite offering professional-grade face generation, editing, and transformation tools for artists, marketers, and casual users. [EIN Presswire]
Voice News
Captello has launched its Meeting Management Platform, integrating lead capture and meeting management to optimize scheduling and boost event ROI through AI-assisted features. [PRWeb]
ElevenLabs has launched v3, a text-to-speech AI model featuring enhanced emotional expression, multilingual capabilities, and human-like vocal characteristics including laughter and whispers. [ZDNet]
And ElevenLabs has restored AFL legend Neale Daniher’s voice through AI technology that cloned his speech patterns from old press conferences, replacing his previous robotic voice. [The Age]
Krisp launched an AI-powered Voice Platform combining noise cancellation, accent conversion, live interpretation, and agent assistance tools for contact centers. [Business Wire]
Rime launched Arcana, a text-to-speech model generating diverse, natural-sounding voices that has boosted sales for companies like Domino’s and Wingstop by 15%. [VentureBeat]
Telnyx launched a Partner Program enabling businesses to scale their offerings through integrated voice, IoT, and AI technologies. [GlobeNewswire]
Voys SA launched AI call transcription services for South African businesses, offering instant, multilingual, searchable records of conversations across multiple industries. [TechCentral]
Document AI
Copyleaks launched a Google Docs Add-on that enables real-time AI content analysis and plagiarism detection directly within users’ documents. [GlobeNewswire]
DeckBird.ai launched version 3.0 of its AI Narrating Agent, transforming static presentations into interactive webinars with AI-powered voices across 73 countries. [Business Wire]
Indico Data launched an AI-powered Agentic Decisioning Platform to help insurance companies automate complex underwriting and claims processes with greater speed and accuracy. [Indico Data]
Liquid Logics has partnered with Lightning Docs to integrate attorney-grade loan document automation into its lending software platform. [EIN Presswire]
Nvidia launched Llama Nemotron Nano VL, a multimodal vision language model that extracts and analyzes data from various document types with high precision. [Nvidia]
Routable launched AI-powered predictive bill coding to automate invoice processing, eliminating manual data entry and learning from user corrections to improve accuracy. [PRWeb]
Turingon launched Proposal Pilot, an AI-powered platform designed to automate federal contract proposal writing while maintaining compliance and security requirements. [Business Wire]
Translation
Apple has expanded its Intelligence features to eight additional languages and introduced AI-powered Live Translation for Messages, Phone calls, and FaceTime conversations. [ZDNet]
And Apple launched multiple Apple Music features, including lyrics translation and pronunciation guides. [TechCrunch]
BookTranslate.ai has outperformed legendary translator Donald Keene in blind tests with its three-stage AI translation system for literary works. [EIN Presswire]
DeepL says the latest Nvidia chips let it translate the whole internet in just 18 days. [CNBC]
Lingopal showcased its sub-200ms real-time AI translation technology at London’s Sports Loft Summit, demonstrating multilingual sports commentary capabilities to industry leaders. [EIN Presswire]
Reddit has translated millions of pages using AI tools, with Google's approval and subsequent documentation updates confirming the acceptability of valuable AI-translated content. [Search Engine Land]
Sanas has launched AI-powered real-time speech-to-speech translation technology that preserves speaker identity while converting between languages, starting with English-Spanish and English-French pairs. [Speech Technology Magazine]
Translated launched Lara V2, an upgraded AI translation system featuring improved accuracy, expanded language support, and enhanced human-AI integration capabilities. [Slator]
Search
Google has begun testing Search Live, an AI Mode feature powered by Project Astra that enables real-time conversations with users through the Google app. [9to5Google]
Google‘s AI tools, including AI Overviews and chatbots, have significantly reduced news publishers’ web traffic, prompting some outlets to pursue content-licensing deals with tech companies. [TechCrunch]
Integris Design launched an AI-optimized citation management platform to help businesses maintain visibility across AI search engines like ChatGPT and Gemini. [EIN Presswire]
Mozilla expressed opposition to Google's plans for integrating Gemini AI into Chrome, fearing it would further entrench Chrome’s market dominance and limit browser competition. [The Register]
Perplexity reached 780 million monthly searches and is preparing to launch Comet, its AI-powered browser, while introducing direct product purchases through PayPal and Venmo. [Yahoo Finance]
ProRata AI has partnered with over 500 publications, including The Atlantic and Fortune, to create one of generative AI’s largest licensed content libraries for its Gist.ai search engine. [Business Wire]
Pureinsights launched Discovery 2.2, integrating Vespa.ai to enable large-scale hybrid search capabilities alongside enhanced security and monitoring features. [EIN Presswire]
SUSO Digital launched an AI SEO service helping agencies optimize client content for AI platforms like ChatGPT through monitoring, optimization, and analytics. [EIN Presswire]
The Browser Company launched Dia, a Mac-exclusive browser featuring an AI chatbot that can analyze users’ browsing data and interact with logged-in websites. [The Verge]
AI in Journalism
IsItCap.com launched a free, ad-free AI platform that analyzes news credibility by comparing mainstream media, alternative sources, and social media sentiment. [EIN Presswire]
Northwestern University’s Local News Initiative published a report exploring AI’s impact on local journalism, highlighting both opportunities for innovation and potential threats to news organizations’ sustainability. [Poynter]
Politico’s AI-powered report generator has produced inaccurate content, including fabricated lobbying efforts and outdated information about Roe v. Wade, prompting complaints from unionized staff. [Semafor]
The New York Times has appointed as AI initiatives director Zach Seward, who described AI as a ‘parlor trick’ with limited practical applications for journalism in the Columbia Journalism Review. [Editor & Publisher Magazine]
Yahoo News launched AI-powered ‘Key Takeaways’ in its relaunched app, joining The Wall Street Journal and Bloomberg in offering automated article summaries to help busy readers. [NiemanLab]
Health Tech
Avio Health has launched an AI platform that transforms disconnected health data into preventive insights by integrating lab results, wearables, and patient histories. [EIN Presswire]
Clinii launched Connect-AI, an in-house communication platform integrating AI-powered documentation and real-time patient interaction tools for chronic care management programs. [Business Wire]
eClinicalWorks has implemented an AI Document Insights Assistant at Ferris Family Medicine, streamlining document processing and improving patient care efficiency. [Business Wire]
G2 Speech launched Aida, an AI-powered digital assistant designed to automate and enhance medical documentation processes for healthcare professionals. [Health Tech World]
Healthy4U launched an AI-powered mobile health app featuring multi-agent intelligence to provide 24/7 personalized care and preventive healthcare recommendations. [EIN Presswire]
Huma Therapeutics launched Hi Scribe, an AI-powered clinical documentation tool that automates note creation across 4,500+ health systems worldwide. [Health Tech World]
IQVIA launched AI agents using Nvidia technology to enhance healthcare workflows and accelerate insights across life sciences and clinical research services. [Business Wire]
Komodo Health launched MapLab Enterprise, a full-stack analytics platform that reduces healthcare insight generation from months to hours while cutting costs by 50%. [Business Wire]
Lumeris has partnered with Google Cloud to enhance its Tom platform with AI capabilities, aiming to address the primary care physician shortage and reduce administrative burdens. [Forbes]
MetaPhy Health launched Cara, an AI nurse that provides personalized patient engagement and support for chronic condition management through automated calls and monitoring. [EIN Presswire]
RadiantGraph launched AI Voice Studio, enabling healthcare organizations to automate HIPAA-compliant voice calls while reducing costs and improving member engagement. [PRWeb]
Viva AI launched an AI-powered dental office platform that handles calls, scheduling, payments, and marketing analytics while providing real-time practice insights. [PRWeb]
Wysa launched Gateway, an AI chatbot that reduced mental health patient intake times by 30 minutes while maintaining 95% diagnostic accuracy in UK trials. [Business Wire]
Zideas AI Lab has partnered with FindYourZen to develop AI-driven mental health solutions targeting Gen Z’s rising depression and anxiety rates. [EIN Presswire]
And the Joint Commission, the US’s oldest health care accreditation organization, has partnered with the Coalition for Health AI to establish guidelines and certification for responsible healthcare AI implementation across US healthcare organizations. [STAT]
Legal Tech
Docusign has expanded its UK partner program with new specializations, tracks, and support to help partners leverage its Intelligent Agreement Management platform. [ITPro]
eimmigration has partnered with Visalaw.ai to integrate AI capabilities into immigration case management, streamlining workflows and enhancing legal outcomes for practitioners. [Business Wire]
England’s High Court ruled that lawyers must verify AI-generated legal research against authoritative sources, following cases where attorneys cited nonexistent precedents. [TechCrunch]
Eve launched a beta version of its AI Intake Specialist, offering law firms 24/7 automated client intake, call support, and case management system integration. [LawSites]
FiscalNote has enhanced its PolicyNote platform with AI-powered features including bill forecasting, redesigned alerts, and similar bill detection to help customers navigate legislative changes. [Business Wire]
Gavel Exec has launched Projects, enabling law firms to train AI using their own documents and preferences without requiring technical expertise or extensive resources. [EIN Presswire]
Nextpoint has partnered with OPVEON to integrate cloud-based litigation software with trial consulting services, enhancing support for legal teams nationwide. [PRWeb]
OpenAI's o3 model has earned grades ranging from A+ to B on law school final exams, significantly outperforming earlier AI versions and matching top students. [Reuters]
Ed Tech
GEMS Education has launched a Global Education AI Hub in Dubai, offering partners access to 90+ schools and 200,000 students for testing AI-driven educational solutions. [EIN Presswire]
StudyPro has launched a free beta version of its all-in-one AI writing platform, featuring integrated tools for content generation, plagiarism detection, and paraphrasing. [EIN Presswire]
Udemy has launched new AI Fluency Packages to help organizations develop AI skills, building on its existing platform that serves 80 million learners globally. [Business Wire]
AI technology is being championed by UK science secretary Peter Kyle to help dyslexic students learn, amid concerns over insufficient human support for those affected. [The Guardian]
Funding
AI Pro University has secured Series A funding and announced a US$1m expansion to develop custom AI education tools and certification programs. [EIN Presswire]
ai.work has emerged from stealth with US$10m seed funding and launched an AI Worker platform that automates enterprise workflows across multiple departments. [Business Wire]
AI4ALL has received a US$2m Google.org grant to launch two initiatives in 2025: the Future of AI Education Council and Future of AI Salon dinner series. [PRWeb]
Altura raised €8m in Series A funding to expand its AI-powered bid management platform into the UK market and introduce autonomous AI agents. [Tech.eu]
Arlequin AI secured €4.4m in seed funding to develop sovereign data operations platforms that combat disinformation using unsupervised AI models. [EU-Startups]
Autonomize AI has raised US$28m in Series A funding to expand its healthcare-focused AI platform that reduces administrative burdens in medical operations. [Business Wire]
Compyl secured US$12m in Series A funding to enhance its AI-powered governance platform and expand operations amid growing market demand. [Yahoo Finance]
Conveyor has launched an AI-powered platform to automate security reviews and RFP processes, securing US$20m Series B funding after attracting clients like Netflix and Zendesk. [TechCrunch]
Definely secured US$30m in Series B funding to expand globally and enhance its AI-powered legal technology platform, bringing total funding to US$40m. [Tech.eu]
Glean raised US$150m in Series F funding at a US$7.2 billion valuation to advance enterprise AI innovation, following rapid growth to US$100m+ annual revenue. [Business Wire]
Lovable, an AI startup enabling non-programmers to create apps, has entered talks with US investors for US$100m+ funding that could value it at US$1.5bn+. [Verdict]
Lyrebird Health raised US$12m for its AI medical transcription platform, which is handling 30,000 daily consultations in Australia. [Startup Daily]
Meta Platforms has agreed to acquire a 49% stake in Scale AI for US$14.8B, with Scale AI’s CEO set to lead Meta’s new superintelligence lab. [Yahoo Finance]
Multiverse Computing raised €189m Series B funding to advance CompactifAI, its quantum-inspired technology that compresses LLMs by up to 95% while maintaining performance. [Tech.eu]
Omnisent raised €2.62M in pre-seed funding to develop acoustic AI technology that converts environmental sound into machine-interpretable intelligence for various industries. [Silicon Canals]
Outset has raised US$17m in Series A funding to expand its AI-powered market research platform, which conducts automated video interviews for Fortune 500 clients. [VentureBeat]
Perscient has completed a US$4.1m funding round and launched an AI-powered narrative intelligence platform that analyzes stories across various media formats. [Business Wire]
Sintra raised US$17m in seed funding after growing from a weekend project to a US$12m ARR business offering AI-powered assistants for small businesses. [Tech.eu]
Speedata has raised US$44m in Series B funding to develop an analytics processor that performed 280 times faster than conventional hardware in testing. [Yahoo Finance]
WeThinkCode secured R35-million from Google.org to train 12,000 people in South Africa and Kenya in AI skills through free programming courses. [TechCentral]
You.com has entered funding talks seeking a $1.4bn valuation, pivoting from general AI search to workplace productivity tools while expanding its enterprise offerings. [Verdict]
Acquisitions
AMD has acquired Untether AI's team and their energy-efficient AI inference chip technology, following its recent purchase of Brium. [TechCrunch]
Newsrooms.ai has acquired Austrian tech media platform Trending Topics, aiming to develop AI-driven content production tools. [Tech.eu]
Salesforce has acquired Moonhub, an AI recruiting startup known for bias-conscious hiring tools, following its US$8B Informatica acquisition as part of its AI-focused expansion strategy. [Yahoo Finance]
Soda has acquired NannyML, integrating AI monitoring capabilities into its data quality platform to enhance issue detection and reduce false alerts across data pipelines. [Silicon Canals]
VENZA has acquired Guest Voice AI, a hospitality platform that automates guest inquiries and service management while integrating with major property management systems. [EIN Presswire]
There’s More
Anthropic has discontinued its ‘Claude Explains’ blog experiment after one month, removing AI-generated posts that had been edited by humans. [TechCrunch]
Google DeepMind CEO Demis Hassabis has predicted humanity will achieve AGI by 2030 and colonize the galaxy. [Futurism]
Here’s Sam Altman’s blog post that foresees widespread, affordable superintelligence by 2030, predicting dramatic advances in AI capabilities while emphasizing safety and equitable distribution. [Sam Altman]
Altman says that each ChatGPT query consumes approximately 0.000085 gallons of water and 0.34 watt-hours of energy. [The Verge]
The Back Cover
Here’s First To Eleven with a cover of Fleetwood Mac’s Dreams.
If you know of a great cover version of a back-catalogue classic, drop an email to us at backcover@language-technology.com and we’ll consider it for inclusion here.
Got this from someone else?
Did you find this newsletter useful? if so, you might forward it to a friend; or you could email us at news@language-technology.com to tell us what you want more of.
Did you hate this newsletter? if so, you could forward it to an enemy; or you could email us at news@language-technology.com to tell us why – make it better!