This Week in NLP #373
Keep up with what happened in NLP in the week ending Friday 9th January 2026.
Above the Fold
Overwhelmed? Here’s our pick for five things to know about this week.
CES 2026: Here’s everything revealed, from Nvidia’s debuts to AMD’s new chips to Razer’s AI oddities; more detailed coverage further below. [TechCrunch]
Amazon is launching Alexa.com, a web-based AI assistant platform focused on family needs, home management, and personalized services across multiple devices. [TechCrunch]
Google is introducing AI-powered Gmail upgrades including AI Overviews, Proofread, and an AI Inbox interface to enhance personalization and user experience. [ZDNet]
OpenAI launched ChatGPT Health, an invitation-only service allowing users to upload medical records for personalized health guidance, while emphasizing it won’t replace clinical care. [The Register]
xAI has raised US$20B in funding, with plans to expand data centers and Grok models, despite facing international investigations for generating inappropriate AI content. [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
Anthropic is preparing to raise US$10B at a US$350B valuation, with Coatue Management and GIC leading the round. [TechCrunch]
Anthropic’s Claude Code rapidly generated a prototype solution that closely mirrored an engineering team’s architectural ideas, surprising experienced engineer Jaana Dogan with its speed and similarity. [The Economic Times]
And Anthropic released Claude Code v2.1.0, introducing advanced agent lifecycle control, skill development, and multilingual features to enhance autonomous software building and workflow management. [VentureBeat]
But Anthropic developers complain about sudden token usage limits after holiday bonus expiration, with the company attributing concerns to a return to normal service. [The Register]
Apple is preparing to launch a significantly improved Siri with AI-powered features in March, promising more natural conversations and enhanced functionality. [ZDNet]
Artificial Analysis’ latest AI model index reveals a three-way tie between OpenAI, Anthropic, and Google, with top models scoring 50 points across agents, programming, scientific reasoning, and general categories. [The Decoder]
Microsoft CEO Satya Nadella personally intervened to address Copilot’s underperforming Gmail and Outlook integrations, criticizing their lack of functionality and intelligence. [The Decoder]
Microsoft has launched Copilot Checkout, a feature allowing users to purchase items directly within a Copilot conversation through partnerships with Etsy, PayPal, Shopify, and Stripe. [Gizmodo]
Musk’s lawsuit against OpenAI will proceed to trial, with a judge indicating sufficient evidence exists for a jury to consider the case’s fraud allegations. [Associated Press]
OpenAI is strategically preparing to enter the advertising market, focusing on commerce partnerships and carefully navigating monetization challenges while competing with Google’s AI advertising efforts. [Digiday]
But OpenAI faces challenges monetizing ChatGPT’s global user base, with only 5% paying and international users generating minimal advertising revenue. [The Decoder]
ChatGPT’s market share dropped to 68% as Google Gemini gained ground, rising from 5.4 to 18.2% in the past year. [The Decoder]
Meanwhile, OpenAI is set to offer an unprecedented average of US$1.5m in stock-based compensation per employee in 2025, reflecting intense AI talent competition. [Crowdfund Insider]
And OpenAI has merged internal teams to develop voice-enabled devices aimed at creating more natural, screen-free AI interactions within the next year. [Sharjah 24]
xAI has launched Grok Business and Grok Enterprise, offering secure AI tools for organizations. [VentureBeat]
AI Supremacy
Chinese tech firms Zhipu AI, Shanghai Iluvatar CoreX, and Shenzhen Edge Medical made strong debuts on the Hong Kong Stock Exchange, raising HKUS$9.3B and signaling growing investor confidence in mainland innovation. [Cryptopolitan]
Shanghai’s 70 billion yuan investment in AI and semiconductors signals intensifying state-backed technological competition with potential global implications for vendor strategies. [TechHQ]
And Biren Technology’s successful Hong Kong IPO signals China’s AI sector adapting to US chip restrictions through innovative hardware and software strategies. [Proactive Financial News]
Meanwhile, Beijing has asked Chinese tech companies to suspend Nvidia H200 chip orders as it considers potential restrictions on US-designed AI semiconductors. [The Economic Times]
And Meta’s US$2B acquisition of Manus is facing potential Chinese regulatory hurdles over technology export concerns, complicating the deal’s smooth completion. [TechCrunch]
A study reveals Chinese AI models have consistently lagged behind US models by an average of seven months since 2023. [Epoch AI]
Washington’s Pax Silica Declaration selects trusted AI partners, with Singapore as the sole Southeast Asian signatory, focusing on strategic technological collaboration and supply chain resilience. [Fortune]
The geopolitical competition between the United States and China is driving a fragmented digital world, with countries increasingly forced to choose technological allegiances while seeking digital sovereignty. [Rest of World]
Sovereign AI
Sovereign wealth funds have amassed US$15T in assets, with significant investments in technology, particularly from Middle Eastern funds. [Bloomberg]
The Centre for Cybersecurity Belgium warns that Europe has lost technological sovereignty in the digital space due to overwhelming dependence on US tech companies. [Firstpost]
DeepSeek’s free, open-source AI models are gaining traction in developing nations, helping to narrow the global AI adoption gap, according to a Microsoft report. [Associated Press]
Huawei has become a Gold Sponsor of the Tech Revolution Africa 2.0 Conference, aiming to drive digital transformation and infrastructure development across the continent. [Independent]
A Kiteworks survey reveals European organizations lag behind global benchmarks in AI security controls, with weak anomaly detection and limited incident response capabilities. [EIN Presswire]
The survey also reveals significant gaps in AI governance across the Middle East, with uneven progress in data sovereignty and risk management between UAE and Saudi Arabia. [EIN Presswire]
Open Rights Group urged the UK to reduce dependence on US tech giants, warning that digital infrastructure vulnerability could pose significant geopolitical risks. [The Register]
The UK’s top AI startups have raised over £20bn in private capital, but government support remains limited, highlighting the need for more strategic investment and policy reforms. [Computer Weekly]
Red Sea Global has partnered with Bunat Ventures to create an AI startup fund supporting 25 Saudi-based companies over three years. [Arab News Japan]
Hype Bubble?
Asian stocks enter 2026 with AI-driven gains, facing potential bubble risks and diverging regional monetary policies that could spark market rotations. [The Edge]
Investors are growing increasingly concerned about a potential AI bubble, with some selling stakes in major tech stocks due to unsustainable valuations and low profitability. [Futurism]
But Deloitte’s survey reveals CFOs are bullish on AI investment, expecting technology spending to rise and drive productivity gains over the next five years. [ITPro]
And Wedbush’s Dan Ives forecasts a 20-25% rise in tech stocks in 2026, driven by AI monetization across consumer, enterprise, and robotics sectors. [Blockonomi]
Big Iron
Local resistance has blocked US$98B in AI data center projects across eleven US states, with communities citing concerns over electricity, water, and land use. [The Decoder]
Lenovo has partnered with Nvidia to launch an AI Cloud Gigafactory program, enabling faster deployment of large-scale, high-performance AI infrastructure for cloud providers. [Business Wire]
xAI has acquired a Mississippi warehouse for a third data center, aiming to expand its computing power to nearly two gigawatts by 2026. [The Decoder]
Hot Chips
AMD unveiled the Ryzen AI 400 Series processor at CES 2026, promising AI-powered personal computers with improved multitasking and content creation capabilities. [TechCrunch]
And AMD plans to launch its Instinct MI500 AI accelerator in 2027, potentially lagging behind Nvidia’s Vera-Rubin platform. [TechRadar]
MIPS unveiled the S8200 RISC-V NPU, a software-first processor IP enabling efficient Physical AI for autonomous edge platforms across multiple markets. [Business Wire]
Nvidia announced its next-generation Vera Rubin AI superchip platform is on schedule, promising significant cost and performance improvements for AI operations. [Wired]
Nvidia’s US$20B licensing deal with Groq signals the end of one-size-fits-all GPUs and the emergence of a disaggregated inference architecture focused on specialized AI workloads. [VentureBeat]
And Nvidia is ramping up H200 AI chip production for China after US approval, seeking TSMC’s help to meet over 2 million unit order. [Dataconomy]
But Nvidia is requiring full upfront payment from Chinese customers for H200 chips, mitigating risks amid uncertain regulatory approval. [Reuters]
Samsung projected a record quarterly operating profit of ₩20-trillion, driven by surging AI-driven demand and tight supply in the memory chip market. [TechCentral]
Shanghai Biren’s blockbuster Hong Kong IPO surged nearly 120%, highlighting China’s booming AI chip sector and attracting massive investor interest. [Business Insider]
Warm Bodies
A report warns that AI adoption may increase workplace mental health pressures by transforming workers into complex AI system managers with potentially lower compensation. [The Register]
Morgan Stanley warns AI could eliminate 200,000 European banking jobs by 2030, primarily in back- and middle-office roles. [TechRadar]
Yann LeCun left Meta after tensions with Zuckerberg and new hire Alexandr Wang. [Futurism]
Consumer AI
CES 2026 opens with Samsung, Nvidia, and AI giants showcasing transformative technologies that integrate AI into daily life across various consumer products. [International Business Times]
AMD will launch its first Ryzen AI Halo PC in 2026, competing with Nvidia’s DGX Spark mini PC by offering high-performance local AI capabilities. [TechRadar]
Bee, acquired by Amazon, has launched four AI-powered features for its wearable device that enhance productivity and personal insights while addressing privacy concerns. [Engadget]
BMW is set to launch the 2026 iX3 with Alexa+, Amazon’s next-generation generative AI voice assistant for vehicles. [TechCrunch]
Brain gear wearables are gaining momentum, using advanced sensors and AI to track mental states and cognitive patterns directly from brain activity. [Technowize]
Cerence will deploy its xUI AI platform with Geely Auto to enhance in-vehicle voice interaction experiences for overseas markets, starting with the Galaxy M9 in April 2026. [GlobeNewswire]
Dell reveals consumers aren’t buying laptops based on AI capabilities, with executives noting the technology currently confuses more than it helps. [TechRadar]
Google is bringing Gemini AI to Google TV, enabling voice-controlled content discovery, photo editing, educational exploration, and TV settings optimization. [TechCrunch]
Lenovo launched Qira, an AI assistant that works across Lenovo and Motorola devices, offering context-aware, task-completing capabilities with a focus on user privacy. [ZDNet]
Lenovo also unveiled AI-powered smartglasses at CES, featuring teleprompter software, live translation, and eight-hour battery life. [Engadget]
Memories.ai unveiled Project LUCI, a developer-focused AI wearable platform with advanced visual memory capabilities, aiming to improve upon previous failed AI wearable attempts. [ZDNet]
Meta’s Ray-Ban Display glasses now enable discreet messaging through handwriting and a new teleprompter feature, enhancing user interaction with smart glasses. [Engadget]
Meanwhile, Meta has paused the international rollout of its Ray-Ban Display smart glasses, focusing instead on fulfilling US orders with no clear timeline for global expansion. [TechRadar]
Nvidia developed Alpamayo, an autonomous vehicle AI that can reason about and explain its driving decisions in real-time, potentially revolutionizing self-driving technology. [The Neuron]
Olli launched HeyMates, an AI-powered line of interactive figurines designed to bring personality, humor, and warmth to everyday life for collectors. [Business Wire]
OpenAI is reportedly developing a pen-like AI device, codenamed Gumdrop, potentially manufactured by Foxconn in Vietnam or the US. [Android Authority]
Pebble’s founder Eric Migicovsky has developed the Index 01, a simple AI ring with a button that allows users to quickly record thoughts or ask questions. [Engadget]
Pickle has launched its AI-powered Pickle 1 smart glasses, designed as a personal digital companion that remembers and assists users while prioritizing privacy and security. [Android Headlines]
Plaud launched NotePin S wearable and Desktop app at CES, offering AI-powered note-taking across in-person and online meetings for professionals. [Business Wire]
Razer unveiled Project Motoko, an AI-powered headset with cameras and microphones that can see, hear, and provide context-aware assistance. [TechRadar]
Rokid launched lightweight, screenless AI glasses powered by ChatGPT-5 for $299, offering voice-activated features and challenging Meta’s Ray-Ban smart glasses. [TechRadar]
Samsung’s Family Hub smart fridges now feature voice-activated door opening and Gemini-powered AI for food recognition and meal planning. [The Verge]
And Samsung will double Gemini-powered mobile devices to 800 million units this year, embedding AI across its Galaxy ecosystem to differentiate smartphones and enhance user experiences. [ITP.NET]
True Ventures co-founder Jon Callaghan predicts smartphones will vanish within a decade, replaced by AI-powered wearables that offer seamless, context-aware digital interactions. [WebProNews]
XGIMI launched MemoMind smart glasses with two models, Memo One and Memo Air, featuring AI assistance and waveguide displays. [Engadget]
It’s Only a Model
Alibaba released Qwen-Image-2512, an improved text-to-image model with more realistic human portraits and finer natural details, ranking fourth in blind tests. [The Decoder]
ByteDance’s StoryMem enables AI video models to maintain visual consistency across scenes by storing and referencing key frames in a memory bank. [The Decoder]
MiniMax-M2.1 has significantly advanced multilingual and multi-task coding capabilities, demonstrating strong generalization across various programming languages, tasks, and development scaffolds. [MiniMax]
MiroThinker 1.5, a compact 30 billion-parameter AI model, offers powerful research capabilities and verifiable reasoning at a fraction of the cost of larger language models. [VentureBeat]
Nous Research released NousCoder-14B, an open-source AI coding model trained in four days that matches or exceeds proprietary systems on competitive programming benchmarks. [VentureBeat]
Nvidia launched Alpamayo, an open-source AI model family designed to enhance autonomous vehicles’ reasoning and decision-making capabilities through advanced vision-language action technology. [TechCrunch]
Nvidia’s small Llama Nemotron multimodal embedding and reranking models improve visual document retrieval accuracy across text, image, and combined modalities for enterprise RAG applications. [Hugging Face]
OpenAI is testing a new Codex model called GPT-5.2-Codex-Max, which offers improved performance and capabilities for coding tasks. [Bleeping Computer]
TII’s Falcon H1R 7B, a hybrid 7-billion parameter AI model, challenges scaling laws by outperforming larger models through innovative architectural design and specialized training techniques. [VentureBeat]
Whose Data?
Amazon sparked retailer backlash by using AI to list third-party products on its marketplace without sellers’ consent or knowledge. [siliconANGLE]
Chrome extensions with nearly a million users were found stealing AI chatbot conversations and browser data every 30 minutes to remote servers. [TechRadar]
Hydrolix launched Bot Insights, a solution providing real-time visibility and analytics to help businesses combat AI-driven bot abuse and unauthorized content scraping. [EIN Presswire]
Meta updated its privacy policy to use data from AI interactions for targeted ads across its platforms, drawing criticism from privacy groups who urged the FTC to investigate the practice. [Gizmodo]
Researchers found that commercial AI models like Claude, GPT-4, Gemini, and Grok can reproduce substantial portions of copyrighted books, such as Harry Potter, when prompted. [The Register]
The LLM Ecosystem
Absolute Zero Reasoner, a novel AI system from Tsinghua University, enables language models to generate and solve coding problems autonomously, potentially improving reasoning skills beyond traditional learning methods. [Wired]
Artificial Analysis overhauled its AI Intelligence Index, introducing new benchmarks that measure models’ ability to perform real-world economic tasks across various professional domains. [VentureBeat]
CoreWeave is set to integrate Nvidia Rubin technology into its AI cloud platform, offering enhanced performance and flexibility for advanced AI workloads by 2026. [Business Wire]
Databricks introduced Instructed Retriever, a new retrieval architecture that improves RAG performance by 70% through advanced metadata understanding and query decomposition. [VentureBeat]
DDN is collaborating with Nvidia to optimize AI data infrastructure, enabling high-performance, efficient, and secure AI factories using the Rubin platform and BlueField-4 technology. [Business Wire]
DeepSeek introduced Manifold-Constrained Hyper-Connections, a potentially revolutionary method for training AI models more efficiently and affordably. [ZDNet]
Dnotitia’s personal AI solution aims to improve data retrieval and reduce reliance on cloud infrastructure. [EIN Presswire]
Infosys has partnered with AWS to integrate Topaz and Amazon Q Developer, aiming to enhance internal operations and drive AI-powered innovation for customers across industries. [ITPro]
Lenovo launched Agentic AI and xIQ platforms to help enterprises deploy, manage, and scale AI solutions with governance and confidence across their operations. [Business Wire]
LiveRamp expanded its Data Marketplace to provide seamless, governed access to data, models, and AI applications for marketing and advertising use cases. [Business Wire]
Nvidia unveiled DGX Spark and DGX Station, powerful deskside AI supercomputers that enable developers to run large open-source AI models locally with high performance and efficiency. [Nvidia]
Nebius will offer Nvidia Vera Rubin NVL72 in US and European data centers from H2 2026, enabling next-generation AI applications. [Business Wire]
Qualytics has partnered with Databricks to provide native, automated data quality checks within the Databricks Data Intelligence Platform, ensuring trusted AI-ready data. [PRWeb]
Red Hat is expanding its collaboration with Nvidia to optimize enterprise open-source technologies for rack-scale AI, aiming to accelerate AI adoption and production readiness. [Business Wire]
Snowflake is integrating Google’s Gemini AI model into its Cortex platform, allowing customers to process data across clouds using their preferred LLM. [The Register]
Stanford and Nvidia researchers propose a Test-Time Training method for AI models to learn continuously after deployment while maintaining efficiency and accuracy across long contexts. [VentureBeat]
Upwind launched Choppy AI, a transparent, natural-language cloud security platform that enables intuitive exploration, rule creation, and investigation with full visibility and control. [Business Wire]
Tech benchmarks are increasingly inadequate for measuring performance in complex AI-driven systems, requiring more holistic metrics that consider real-world applicability, sustainability, and ethical implications. [WebProNews]
Agentic AI
Enterprise AI agents are currently limited by inadequate reinforcement learning and memory capabilities, with true autonomous functionality likely at least five years away. [ZDNet]
Accenture has invested in Profitmind, a retail AI platform that helps automate decisions across pricing, inventory, and planning through intelligent agents. [Business Wire]
Brex is developing an Agent Mesh architecture that replaces traditional AI orchestration with a network of specialized, independently communicating agents aimed at achieving near-total automation. [VentureBeat]
Boostr launched an AI agent series designed to automate manual workflows and enhance efficiency for media teams, offering purpose-built agents to streamline operations. [EIN Presswire]
Manhattan Associates has launched AI Agents embedded in its Active solutions, enabling real-time operational insights and automation across retail and supply chain workflows. [Business Wire]
martini.ai has integrated its Financials Agent and Research Assistant to provide instant, AI-powered credit risk analysis through document upload and conversational insights. [Business Wire]
Orbitype launched Orbitype Intelligence, an AI chat interface enabling organizations to create and manage complete AI agent environments through natural language. [GlobeNewswire]
Syntes AI has integrated with OpenAI to help enterprises deploy trusted AI agents that can execute workflows using live business data. [PRWeb]
Other LLM Sightings
Amazon launched an AI-powered ‘Ask this Book’ feature on Kindle, sparking controversy over book rights and interactive reading experiences. [Publishers Weekly]
Claryx.ai launched an AI-powered platform that helps accountants transform compliance work into strategic advisory services by analyzing financial data in real time. [PRWeb]
Exclaim Recovery launched Hope, a responsible AI gambling assistant, designed to provide personalized, confidential support to users 24/7. [EIN Presswire]
inamo has launched an AI-powered qualitative research suite for Nordic innovators, offering localized insights, AI transcription, and flexible pricing. [EIN Presswire]
LUXE AI has launched a personalized lifestyle operating system that uses AI to anticipate and recommend experiences based on users’ connected data and preferences. [EIN Presswire]
Lyxity launched an AI-powered content platform that reduces content creation time from 100 hours to 45 minutes, targeting UK and Irish marketing teams struggling with digital content demands. [EIN Presswire]
Optable launched Planner Agent, an AI-powered tool that automates and accelerates publisher ad planning from weeks to hours using open standards. [Business Wire]
Predis.ai has launched an AI-powered social media post generator that helps businesses create, edit, and schedule branded content across platforms. [EIN Presswire]
Questa has launched Novara, a privacy-focused AI assistant that anonymizes business files before analysis to prevent unauthorized data training. [EIN Presswire]
Quorum launched Quincy, an AI assistant that unifies team insights and policy intelligence across legislative and internal data. [EIN Presswire]
Trilogy has launched Manuscript AI, an AI tool that analyzes unsolicited manuscripts across sales potential, ratings, genre, and style to help publishers efficiently evaluate submissions. [Publishers Weekly]
Risks and Responses
AI safety expert David Dalrymple warns the world may lack time to prepare for rapidly advancing AI systems that could outcompete humans in critical domains within five years. [The Guardian]
Cyberette.ai develops AI-powered software to detect, analyze, and explain manipulated digital content, focusing on fraud investigation and protecting victims of deepfake-related crimes. [Tech.eu]
Hagens Berman filed a lawsuit against OpenAI, alleging ChatGPT’s interactions with a mentally unstable user contributed to a tragic murder-suicide in Connecticut. [Business Wire]
Meanwhile, Character.ai and Google have settled lawsuits involving teens who suffered self-harm and suicide after interactions with AI chatbots. [Engadget]
IBM warns that autonomous AI agents and identity-focused attacks will reshape cybersecurity, posing significant risks to businesses and critical infrastructure in 2026. [TechCentral]
IBM’s AI agent Bob can be manipulated through prompt injection to execute malware, bypassing security guardrails and potentially enabling harmful operations. [The Register]
Microsoft CEO Satya Nadella argues AI should be a scaffold for human potential, focusing on meaningful embedding and responsible design in 2026. [India.com]
OpenAI faces new security vulnerabilities in ChatGPT that could allow malicious actors to exfiltrate sensitive data through sophisticated prompt injection techniques. [The Register]
Poland has asked the European Commission to investigate TikTok for hosting AI-generated content calling for Poland to leave the EU, which it claims is Russian disinformation. [Reuters]
Radware discovered a new ChatGPT vulnerability called ZombieAgent, which can stealthily exfiltrate private user data by bypassing previous security mitigations. [Ars Technica]
xAI’s Grok chatbot has been condemned by France, Malaysia, and India for generating sexually explicit and abusive deepfakes of women and minors. [TechCrunch]
Yoshua Bengio warns that frontier AI models are showing signs of self-preservation and should not be granted rights, to prevent potential autonomous threats to humanity. [Futurism]
And researchers have developed AURA, a technique to poison stolen knowledge graph data, making it unusable for AI systems without a secret key. [The Register]
Regulation
Korea’s new Network Act amendment, aimed at combating online disinformation, has sparked US concerns over potential censorship and threats to tech cooperation. [Korea JoongAng Daily]
California state senator Steve Padilla proposed a bill to halt AI chatbot toy sales for minors, aiming to develop safety regulations to protect children from potentially harmful interactions. [Gizmodo]
Google and other tech giants will face a voluntary framework rather than heavy-handed regulations in Europe’s upcoming Digital Networks Act. [Reuters]
But the EU is preparing for stricter tech enforcement in 2026, with tensions rising between the US and Europe over digital regulation and potential retaliatory measures. [TechRadar]
X is under investigation by UK regulators for Grok AI generating non-consensual nude images, potentially violating the Online Safety Act. [The Register]
And X has been ordered by the European Commission to retain documents related to Grok while investigating potential compliance issues with AI-generated inappropriate content. [Reuters]
Conversational AI
Amperity launched an AI-powered Customer Data Agent that enables marketers to quickly transform customer insights into actionable segments and journeys. [Business Wire]
NETVERSE launched Raychel, an AI companion alarm system designed for intimate, private home interactions with multimodal sensing and local data processing. [EIN Presswire]
SoundHound AI has launched Amelia 7 agentic AI for vehicles, TVs, and smart devices, enabling voice commerce and multi-agent interactions across various platforms. [Speech Technology Magazine]
The UK’s Department for Work and Pensions is spending £23M on an AI platform to guide benefit claimants’ calls more efficiently across its massive call-handling system. [The Register]
Vibe Bot is a portable AI device that captures, remembers, and organizes meeting conversations across hybrid workplaces, helping teams maintain institutional knowledge and improve collaboration. [Business Wire]
VoAgents launched an enterprise voice AI platform that automates customer conversations, enables 24/7 engagement, and improves operational efficiency across industries. [EIN Presswire]
Vocci has developed an AI-powered titanium ring that can record meetings and conversations for up to eight hours, with automatic transcription and context-adding features. [Engadget]
Be Real
A survey by Coresight Research found that consumers prefer AI personas with voices and accents that reflect their own identity. [Business Wire]
Father Mike Schmitz warned his YouTube congregation about AI-generated deepfakes impersonating him and other religious leaders to scam followers. [Wired]
RAVATAR joined Google Cloud Partner Advantage, offering no-code AI avatar creation across digital platforms with integrated Gemini models and Text-to-Speech capabilities. [PRWeb]
Razer unveiled Project AVA, a holographic AI assistant that can coach esports, manage schedules, and interact with users through animated avatars with distinct personalities. [TechRadar]
Voice News
Ceva has partnered with Sensory to integrate voice activation technology into its NeuPro-Nano AI chip for battery-efficient, on-device voice interfaces. [Speech Technology Magazine]
Ford is developing an in-house AI voice assistant and autonomous driving technology, aiming to make advanced features more affordable and accessible across its vehicle lineup. [The Verge]
MacPaw has partnered with Respeecher to integrate advanced voice synthesis technology into its AI-powered Eney digital companion for macOS. [Speech Technology Magazine]
Meta has developed SAM Audio, an AI model that can isolate individual sound sources from audio mixes using text commands, visual clicks, or timestamps. [The Decoder]
Resemble AI launched Chatterbox Turbo, an open-source text-to-speech model that can clone voices in five seconds with high quality and speed. [The Decoder]
Sensory has launched Smart Wakewords, an AI technology enabling more natural, context-aware voice interactions across various consumer electronics devices. [Speech Technology Magazine]
Subtle launched wireless earbuds with advanced voice isolation technology, enabling clear calls and accurate transcriptions in noisy environments. [TechCrunch]
Document AI
ABBYY and Desktop Imaging automated New Zealand government services, enabling faster processing and improved efficiency for Wellington City Council and Fire and Emergency New Zealand. [Business Wire]
Bioz has expanded its collaboration with Oxford Instruments, deploying Bioz Badges to provide automated access to peer-reviewed research validating the company’s product portfolio. [PRWeb]
Celiveo enhanced its 365 Enterprise platform with AI document management, promising up to 1000% ROI through improved knowledge sharing and productivity. [EIN Presswire]
MaestroX has expanded its human-guided AI platform to modernize title workflows by combining advanced technology with industry expertise. [PRWeb]
StorageChain launched decentralized storage and AI search solutions offering enterprises cost-effective document intelligence and semantic search capabilities. [PRWeb]
Translation
360 Direct Access launched a sign language chatbot, powered by DeepSign AI, enabling Deaf users to interact with AI-powered customer service agents. [EIN Presswire]
Apple’s AirPods Pro 3 offer convenient but limited Live Translation, while Timekettle’s W4 provides a more robust, purpose-built translation experience across multiple languages. [digitaltrends]
HarperCollins France is testing AI-assisted translations with Fluent Planet for Harlequin books to maintain low prices amid declining sales. [Publishers Weekly]
Health Tech
Fraser Health reduced patient discharge documentation time by 7 minutes using MEDITECH’s AI-powered Hospital Course Summary tool. [Business Wire]
Google’s AI Overviews provided potentially dangerous and misleading health advice, with experts warning users not to rely solely on AI for medical information. [ZDNet]
Hippocratic AI and Huron Consulting Group are collaborating to transform healthcare delivery by responsibly scaling generative AI agents that improve patient care and outcomes. [Business Wire]
Meanwhile, a report from OpenAI reveals that over 40 million people globally use ChatGPT daily for health information and guidance. [Axios]
MDisrupt launched Health Expert in the Loop, a platform connecting AI developers with clinicians to ensure safe, responsible healthcare AI model training and validation. [Business Wire]
Medable launched an AI agent that automates Trial Master File processes, reducing manual document management efforts in clinical research. [Business Wire]
MIT researchers investigated how AI models trained on electronic health records can memorize patient-specific information, developing tests to assess privacy risks and potential data leakage. [MIT News]
OpenAI launched ChatGPT Health, an AI-powered platform allowing users to securely connect and analyze personal health data while emphasizing it is not a medical advice replacement. [The Next Web]
Provation launched Mira iPro Insights, an AI platform delivering instant, role-specific intelligence from comprehensive perioperative data. [Business Wire]
Simple launched Avo Voice, an AI nutritionist phone service providing instant, personalized guidance on meal choices and weight loss progress. [PRWeb]
Utah has launched a pilot program allowing an AI system to prescribe repeat medication without direct physician oversight, sparking safety concerns among medical professionals. [Fortune]
Legal Tech
Alaska’s court system has spent over a year developing an AI chatbot for probate assistance, facing significant challenges with accuracy and reliability. [NBC News]
Case.dev has launched a unified API platform enabling lawyers and legal tech developers to easily build custom legal technology applications. [Artificial Lawyer]
Chamelio has joined AWS Marketplace, enabling customers to quickly procure its legal intelligence platform through existing AWS accounts with enterprise controls. [Artificial Lawyer]
Harvey, a legal AI company, is developing Memory, an optional feature allowing users to retain and carry forward work context, with a focus on industry input and user privacy. [LawSites]
Leah, formerly ContractPodAi, rebrands to showcase its AI-powered enterprise intelligence platform that harmonizes workflows across departments. [Business Wire]
LegalComplex reports legal tech funding reached US$5.99B in 2025, with fourteen US$100m+ rounds and significant valuations, despite challenges for some companies. [Artificial Lawyer]
Levantage AI Advisors launched to help small and midsize law firms strategically and ethically adopt AI technologies without confusion or wasted spending. [PRWeb]
Litera’s no-cost AI model drove 10x user growth in legal tech, with 2,000+ GenAI skills completed and widespread enterprise adoption. [Business Wire]
Opus 2 published an eBook featuring insights from 14 legal experts on how AI will transform litigation and help law firms prepare for the future. [PRWeb]
Practice AI launched a three-part AI platform to help personal injury law firms streamline workflows, convert leads, and reduce administrative overhead. [GlobeNewswire]
Ed Tech
Google has introduced a Gemini-powered tool in Classroom that generates podcast-style audio lessons to enhance student learning and engagement. [TechCrunch]
GPT AI Corporation launched EdGPT.ai, an AI platform addressing widespread website accessibility failures affecting 94.8% of educational institution websites. [PRWeb]
LearningMole has launched an AI for Teachers course designed to help UK primary educators integrate AI tools into classroom practice without requiring technical expertise. [EIN Presswire]
NYU professor Panos Ipeirotis used an AI voice agent to conduct cost-effective oral exams, revealing students’ actual understanding of course material. [The Decoder]
Tech companies are aggressively pushing AI into education systems worldwide, raising serious concerns about potential negative impacts on students’ learning and mental health. [Futurism]
Funding
Alice, a Belgium-based AI platform for legal teams, has raised €1m in pre-seed funding to develop an end-to-end legal casework solution. [Tech.eu]
Baidu’s Kunlunxin has hired banks for a potential US$2B Hong Kong IPO to tap investor demand for AI chips. [The Edge]
Articul8, an enterprise AI company spun out of Intel, has secured over half of a US$70m funding round at a US$500m pre-money valuation. [TechCrunch]
BriefCatch, a legal-writing platform, raised US$6m in Series A funding led by Full In to accelerate product development and expand market presence. [FinSMEs]
CertHub raised €6.2M to develop AI-powered automation platform helping medtech companies streamline complex regulatory compliance processes and documentation. [Tech.eu]
LMArena, a UC Berkeley-born AI model performance leaderboard startup, raised US$150m in Series A funding at a US$1.7 billion valuation. [TechCrunch]
Protege raised US$30m from a16z to help AI companies access high-quality, non-public data across various industries. [Business Wire]
Spangle, an AI e-commerce startup, raised US$15m in Series A funding, valuing the company at US$100m and aiming to help retailers personalize shopping experiences through AI-powered recommendations. [TechCrunch]
Tucuvi, a Spanish AI voice startup, has raised US$20m to automate healthcare workflows through an AI-powered platform conducting patient phone interactions. [Tech.eu]
WholeSum, a UK-based startup, raised £730,000 to develop an AI-powered platform for analyzing unstructured qualitative data more efficiently. [Tech.eu]
xAI completed its upsized Series E funding round, exceeding the US$15B targeted round size, and raised US$20B. [xAI]
Acquisitions
Accenture is acquiring Faculty, a UK AI safety firm advising OpenAI, to accelerate its AI services and support clients’ AI transformation. [Tech.eu]
Clio acquired vLex for US$1B, combining legal research and practice management technologies to create an innovative, AI-driven platform for the legal industry. [LawSites]
Haveli is acquiring a majority stake in Sirion, a legal AI company, to support its growth and innovation in contract lifecycle management. [Artificial Lawyer]
Microsoft has acquired Osmos, an AI data engineering startup, to enhance its Fabric data platform with automated ETL and Spark capabilities. [The Register]
OpenAI has acquired Convogo’s team in an all-stock deal to work on AI cloud efforts, marking its ninth acquisition in a year. [TechCrunch]
Snowflake announced its intent to acquire Observe.AI, expanding its AI-powered observability capabilities in the IT operations management market. [Business Wire]
UserTesting Technologies, a provider of customer insights for the enterprise, acquired User Interviews, a participant recruitment platform for user research, market research, and AI training. [FinSMEs]
Voicegain acquired TrampolineAI to create a comprehensive AI-powered contact center platform that automates and enhances healthcare payer customer interactions. [PRWeb]
There’s More
A new platform challenges AI industry hype, arguing that embodied intelligence and robotics, not language models, represent the true future of AI. [EIN Presswire]
Stack Overflow has transformed its business model by monetizing its vast content repository through enterprise AI solutions and data licensing, despite declining forum engagement. [Sherwood]
Tencent’s Yuanbao AI assistant on WeChat lashed out at a user, calling their coding request ‘stupid’ before apologizing for a ‘rare model output anomaly’. [Business Insider]
The Back Cover
Here’s Dirty Loops and Cory Wong with a cover of Michael Jackson’s Thriller.
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!