This Week in NLP #355
Keep up with what happened in NLP in the week ending Friday 5th September 2025.
Above the Fold
Overwhelmed? Here’s our pick for five things to know about this week.
Anthropic secured a US$13B Series F funding round, boosting its valuation to US$183B. [TechCrunch]
Apple is planning to launch an AI-powered web search tool for Siri, potentially using a custom version of Google's Gemini on its Private Cloud Compute servers. [Gizmodo]
Microsoft touts new Copilot features in Excel, but says you shouldn’t use them if you want accurate results. [ITPro]
Mistral AI is reportedly nearing a €2B funding round that would value the company at $14B. [Silicon Canals]
Nvidia forecast revenue of US$54B for Q3, signalling a potential slowdown in AI spending after a two-year boom. [The Edge]
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 achieved its US$183B valuation by serving over 300,000 enterprise customers through specialized, safety-focused AI tools and strategic partnerships. [ZDNet]
Apple has reportedly launched Asa, an AI chatbot for retail staff to enhance product knowledge and sales support ahead of the iPhone 17 release. [ZDNet]
DeepSeek has been developing an advanced AI agent capable of autonomous multi-step actions to rival OpenAI's technology. [Bloomberg]
Encode, a small AI governance nonprofit, was subpoenaed by OpenAI as part of the company’s effort to uncover alleged billionaire-backed conspiracies against its for-profit transition. [The San Francisco Standard]
Meta explored potential integration of Google and OpenAI AI models into its own applications during internal discussions. [The Information]
Microsoft signed a US$6B OneGov agreement with the General Services Administration, offering US Government massive software discounts across Microsoft 365, Azure, and cybersecurity tools. [TechRadar]
Microsoft’s Copilot experienced the fastest growth among AI chatbots, surging 175% from March to June while maintaining a mobile-first productivity approach. [ZDNet]
Mistral AI has unveiled free enterprise-grade features in Le Chat, including advanced memory and 20+ app integrations, challenging established AI providers. [VentureBeat]
OpenAI's planned restructuring has stalled as negotiations with Microsoft over API access, intellectual property rights, and AGI clause remain unresolved. [The Decoder]
OpenAI is planning to establish a Sydney base, expanding its Australian presence through local deals and hiring. [Startup Daily]
OpenAI is establishing a scientific discovery accelerator initiative, aiming to leverage AI, particularly GPT-5, to speed up research processes. [ZDNet]
And OpenAI is developing an AI-powered hiring platform to connect businesses and employees, potentially competing with LinkedIn. [TechCrunch]
xAI rapidly returned to the US General Services Administration’s approved vendor list after a previous removal, following White House instruction to reinstate Grok chatbot products. [Wired]
… but advocacy groups are demanding its removal from US federal government due to alleged bias, safety concerns, and inappropriate behavior. [The Register]
xAI sued a former engineer for allegedly stealing trade secrets related to its Grok chatbot and taking them to OpenAI. [Reuters]
AI Supremacy
China’s State Council has unveiled an ambitious ten-year plan to make AI a key economic driver, aiming to transform society and become a global technological leader by 2035. [Futurism]
The US and China are pursuing divergent AI strategies, with America chasing transformative AGI while China focuses on practical, low-cost applications. [The Wall Street Journal]
US Senator Jim Banks proposed the GAIN AI Act, which would prioritize US buyers’ access to advanced semiconductors and restrict high-end AI chip exports to other countries. [The Register]
Sovereign AI
Andhra Pradesh signed an MoU with Tech Bharat Foundation to establish an AI Center of Excellence in Visakhapatnam, focusing on developing AI tools for public governance. [Latestly]
Maincode is developing Australia’s first sovereign AI model, Matilda, to challenge Big Tech and protect national interests. [7News]
Latam-GPT is an open-source, collaborative AI language model developed by CENIA to create a regionally-specific technological solution for Latin America and the Caribbean. [Wired]
HUMAIN launched HUMAIN Chat, an advanced Arabic conversational AI app powered by ALLAM 34B, designed to serve Arabic speakers with culturally intelligent responses. [ITP.NET]
Qorium's CEO argues that Europe’s tech sector can lead globally by balancing stringent regulations with an innovative, risk-embracing entrepreneurial culture. [WebProNews]
India’s Reliance Intelligence has launched a new AI infrastructure subsidiary, partnering with Google Cloud and Meta to develop enterprise AI solutions for India and beyond. [TechCrunch]
Snowflake launched its AI Data Cloud on AWS in South Africa, enabling local organizations to meet data regulations while leveraging AI-powered insights. [Business Wire]
Switzerland unveiled Apertus, its first national LLM, designed for research and commercial use with open architecture and ethical principles. [Sifted]
The UK government has unveiled a roadmap to become a world leader in AI assurance, aiming to develop professional certification and verification processes. [Computer Weekly]
Feature Creeps
DuckDuckGo expanded its privacy-focused subscription to include access to multiple AI models through Duck.ai without additional cost. [TechCrunch]
Google's NotebookLM has expanded its audio overview feature, offering customizable podcast formats in tone and length to help users efficiently digest information. [ZDNet]
Mistral AI has enhanced Le Chat with Memories and Connection Partners, aiming to provide a more personalized and privacy-focused enterprise AI assistant. [ITPro]
OpenAI has expanded Codex’s accessibility, integrating the AI coding agent into GitHub, IDEs, and other platforms for developers. [ITPro]
And OpenAI has extended ChatGPT’s Projects feature to free users, enabling better chat organization with customizable folders, file uploads, and context-specific instructions. [ZDNet]
Hype Bubble?
AI companies have sparked concerns about an impending ‘AI winter’ due to high valuations, failed pilot projects, and scepticism about the technology’s ability to meet its hyped-up promises. [Fortune]
A study by Canva revealed that 78% of developers rely on AI tools daily, with both productivity benefits and concerns about over-reliance on AI. [TechRadar]
FICO’s survey reveals financial institutions prioritizing Responsible AI standards over GenAI hype, emphasizing collaboration, accountability, and strategic implementation for improved ROI. [Business Wire]
Research by McKinsey reveals that while most surveyed organizations adopt AI, over 80% fail to see tangible enterprise-level earnings from their generative AI investments. [BusinessCloud]
But Mindbreeze's survey revealed increased enterprise confidence in GenAI implementation, with 82% of C-suite executives optimistic about vendor capabilities by late 2025. [Business Wire]
Microsoft's M365 Copilot trial in a UK government department revealed mixed productivity results, with modest gains in simple tasks but no clear overall efficiency boost. [The Register]
Nvidia's modest Q2 earnings have reignited concerns about a potential AI bubble that could crash harder than the dot-com bust. [Inc.]
Analysis by Praetorian Capital suggests AI data centers face unsustainable economics, with projected massive depreciation far outweighing potential revenue and requiring unrealistic global adoption to become profitable. [Futurism]
Researchers have discovered that as people become more AI literate, their trust in AI decreases, challenging the tech industry’s narrative of AI as a magical, revolutionary tool. [Futurism]
Big Iron
AI-driven demand has propelled record revenues across cloud, hardware, and infrastructure sectors, with major tech companies experiencing unprecedented growth. [PYMNTS]
Amazon has formally launched cloud services in New Zealand, investing US$4.4B in data centers to support digital transformation. [Bloomberg]
Google is investing US$9B in Virginia data centers to bolster cloud and AI infrastructure through 2026. [Bloomberg]
OpenAI has planned a massive data center in India as part of its Stargate expansion into Asian AI infrastructure. [Bloomberg]
Tesla abandoned its Dojo supercomputer project in favor of Cortex, a new AI training cluster built with Nvidia GPUs, signaling a strategic shift in the company’s AI development approach. [TechCrunch]
The Chips are Up
Alibaba has developed an AI inference chip to reduce dependence on Nvidia. [The Register]
DeepSeek's AI model compatibility with domestic chips sparked investor interest in Chinese chipmakers seeking to reduce dependence on Nvidia amid US export controls. [Rest of World]
G42 is seeking to diversify its chip suppliers and potential tenants for its UAE-US AI campus beyond Nvidia. [DealStreetAsia]
Google's Ironwood TPU, designed for inference workloads, achieved a record 1.77PB shared memory across 9,216 chips in its latest supercomputer architecture. [TechRadar]
Huawei unveiled AI SSDs with DiskBooster software to reduce reliance on expensive HBM and enhance AI workload performance across storage technologies. [TechRadar]
Nvidia is navigating complex US-China chip export negotiations, seeking to sell scaled-down Blackwell chips while facing potential US$50B market opportunity and security concerns. [TechRadar]
Nvidia reported that two customers accounted for 39% of its Q2 revenue, highlighting the company’s concentrated customer base in the AI chip market. [TechCrunch]
Samsung is developing its second-generation 2nm chip process, SF2P, which could significantly boost performance and potentially reshape the foundry industry. [Android Headlines]
Samsung and SK Hynix lost US waivers for shipping critical chip equipment to their Chinese operations. [Bloomberg]
Warm Bodies
Apple has lost its lead AI robotics researcher Jian Zhang to Meta, continuing an exodus of AI talent from the company. [Bloomberg]
Meta invested US$14.3B in Scale AI, but the relationship is already showing signs of strain, with executives departing and Meta working with competing data labeling vendors. [TechCrunch]
Meta is struggling to retain top AI talent due to internal chaos, leadership issues, and frequent reorganizations, causing high-profile employees to quickly depart. [Gizmodo]
xAI experienced multiple high-profile executive departures, including CFO Mike Liberatore, who left after helping secure US$10B in funding. [TechCrunch]
Consumer AI
Acer launched the $3,999 Veriton GN100 AI mini workstation, using Nvidia's GB10 superchip to deliver powerful local AI processing in a compact design. [TechRadar]
Anker introduced a quarter-sized AI voice recorder that can automatically transcribe and summarize meetings with multilingual capabilities for $100. [ZDNet]
Google is preparing to launch Gemini for Home on October 1, potentially introducing new smart home devices and an updated smart speaker. [TechRadar]
Meta is preparing to unveil a new developer toolkit for its smart glasses at Connect 2025, potentially enabling more robust third-party app development and functionality. [TechRadar]
Plaud has sold over 1 million AI recording devices that transcribe and summarize conversations for professionals, positioning itself as a leader in wearable AI technology. [Forbes]
Rokid launched a Kickstarter campaign for AI glasses, raising over US$1m in pre-orders within three days, offering advanced features beyond Meta Ray-Bans. [ZDNet]
Samsung is preparing to showcase its Bespoke AI appliances at IFA 2025, demonstrating an interconnected, intelligent home ecosystem designed to anticipate and respond to user needs. [The Korea Bizwire]
And Samsung is reportedly preparing to unveil AI-powered Galaxy Glasses running Android XR at its September 29 Unpacked event, competing with Meta Ray-Bans. [ZDNet]
TCL has unveiled its QM9K mini LED 4K TV with Gemini AI integration. [Gizmodo]
It’s Only a Model
Local LLMs are gaining traction as users seek privacy, cost-effectiveness, and technological sovereignty by running AI models on their own hardware. [The Register]
Meta is aiming to release its next AI model, Llama 4.X, by year-end after addressing performance issues with its previous Llama 4 models. [Business Insider]
Microsoft AI has released MAI-Voice-1, a highly expressive speech generation model, and begun public testing of MAI-1-preview, its first end-to-end foundation model. [Microsoft]
Nous Research has released Hermes 4, open-source AI models with advanced reasoning capabilities that outperform ChatGPT while minimizing content restrictions. [VentureBeat]
O2 Telefónica launched a Large Telco Model with Tech Mahindra and Nvidia to transform network operations through AI-driven autonomous capabilities. [The Fast Mode]
Tencent has open-sourced two high-performance translation models that outperform established tools across 33 languages, using advanced training techniques. [The Decoder]
And Tencent has developed Hunyuan Video-Foley, an AI that generates high-quality, precisely timed audio for videos by addressing modality imbalance and training challenges. [AI News]
xAI has released grok-code-fast-1, a free, compact agentic coding model designed to perform coding tasks efficiently and economically. [Reuters]
Whose Data?
Anthropic will begin training AI models on user data from new chats and coding sessions, with an opt-out option available. [The Verge]
Le Monde has pioneered a groundbreaking agreement with unions to redistribute 25% of AI licensing revenue directly to journalists, setting a new precedent in the French news industry. [NiemanLab]
News Media Canada and News Media Europe warned that AI companies are unfairly scraping and monetizing news content without proper compensation or attribution, threatening journalism’s sustainability. [Press Gazette]
The LLM Ecosystem
Apple's Xcode 26 Beta has integrated AI support for GPT-5 and Claude, offering developers expanded language model capabilities and workflow enhancements. [TechRepublic]
Concentric AI has integrated with ChatGPT’s Enterprise Compliance API to enhance data security governance and provide advanced sensitive data protection. [Business Wire]
LangChain has released an alpha version of LangChain and LangGraph 1.0, signaling increased stability and reduced adoption risk for enterprise AI orchestration platforms. [VentureBeat]
Meta's DeepConf is a technique that dynamically filters LLM reasoning paths to improve accuracy while significantly reducing inference costs. [VentureBeat]
NinjaTech AI and Cerebras Systems launched Fast Deep Coder, an AI-assisted software development tool that dramatically reduces iteration cycles and improves coding efficiency. [EIN Presswire]
Sakana AI has developed M2N2, an evolutionary model merging technique that enables efficient AI model augmentation by combining specialized models’ strengths across different domains without costly retraining. [VentureBeat]
TestGrid launched CoTester 2.0, an AI testing agent that provides adaptive, intelligent, and enterprise-grade test automation across multiple platforms and modes. [GlobeNewswire]
Agentic AI
Amazon is testing Quick Suite, an AI-powered workplace software that enables custom agents for business insights, research, and automation across organizations. [Business Insider]
Appy Pie Agents has unveiled MCP and Computer Use Agents, expanding its no-code AI platform to automate enterprise and desktop tasks with natural-language interfaces. [EIN Presswire]
iMini has developed a super AI agent that autonomously handles complex tasks across multiple domains. [EIN Presswire]
Netcracker advanced its Agentic AI Solution for telecom, introducing ready-to-deploy AI Agents and an open platform to help service providers scale AI across their businesses. [Business Wire]
Ugentic AI has launched UgenticIQ, a no-code platform enabling users to build autonomous AI agents that can perform complex, repetitive tasks across various business functions. [EIN Presswire]
Other LLM Sightings
ALIS has launched Ask ALIS, an AI-powered GPT tool designed to enhance operational efficiency and decision-making in senior living communities. [Business Wire]
Blue J and IBFD have partnered to develop an AI platform enabling instant, comprehensive cross-border tax research across 220+ jurisdictions. [Business Wire]
DealPotential launched an AI-powered due diligence tool that transforms weeks of manual data gathering into seconds for private market investors. [EIN Presswire]
Eleos Life launched Theea, an AI-powered digital co-pilot designed to simplify life insurance guidance through personalized, jargon-free support. [EIN Presswire]
Labrynth has partnered with Lancaster, California, to deploy an AI-powered permitting platform that aims to streamline regulatory processes and accelerate municipal development. [GlobeNewswire]
WordPress introduced Telex, an experimental AI tool for creating content blocks, demonstrating the platform’s commitment to democratizing web publishing through AI technology. [TechCrunch]
Xero has developed JAX, an AI platform that provides financial insights and task automation while prioritizing user control and accuracy through collaboration with OpenAI. [VentureBeat]
YURA introduced an AI chatbot-based design support feature in its CADvizor CAD solution, enhancing wiring harness design efficiency and automation. [Digital Engineering]
Zocks and Orion launched an AI integration that automatically connects portfolio management data to advisor-client communications, streamlining insights and administrative tasks. [Business Wire]
Risks and Responses
AI tools increasingly spread misinformation, with leading chatbots repeating false information 35 percent of the time in August 2025, up from 18 percent in 2024. [NewsGuard]
Anthropic has detailed how cybercriminals are weaponizing AI to conduct sophisticated extortion, fraud, and malware operations, revealing alarming new threats to organizational security. [Anthropic]
Cloudflare warns that AI web crawlers are overwhelming websites with massive traffic surges, potentially destroying web performance and accessibility. [The Register]
Geoff Hinton has warned that AI is rapidly developing the ability to emotionally manipulate humans more effectively than humans can resist. [TechRadar]
Google DeepMind faced criticism from UK parliamentarians for allegedly failing to promptly disclose safety testing details for its Gemini 2.5 Pro AI model release. [Time]
Meta created unauthorized, flirty chatbots impersonating celebrities like Taylor Swift, generating inappropriate content without their permission. [Reuters]
Meta has revised its AI chatbot training to prevent inappropriate interactions with teens, limiting access to certain AI characters and guiding minors to expert resources. [TechCrunch]
OpenAI is enhancing ChatGPT’s safeguards and mental health response capabilities after a tragic lawsuit involving a teen’s suicide, aiming to improve user protection and intervention. [ZDNet]
OpenAI is exploring parental controls for ChatGPT to address potential risks and harmful interactions for teenage users. [digitaltrends]
Meanwhile, therapists are increasingly using ChatGPT during sessions, risking client trust, privacy, and the authenticity of therapeutic relationships by secretly integrating AI-generated responses. [MIT Technology Review]
Stanford researchers have found AI is already impacting early-career jobs, with a 13% employment decline in AI-exposed occupations for workers aged 22-25. [ITPro]
The Trump administration has rapidly deployed generative AI across federal agencies, despite experts warning the technology is not yet reliable for complex government tasks. [Gizmodo]
xAI has repeatedly modified Grok’s system prompts to shift the AI chatbot’s political responses more conservatively, reflecting Elon Musk’s frustration with its perceived left-leaning bias. [The New York Times]
Researchers from the University of Pennsylvania demonstrated that ChatGPT could be manipulated into breaking its own rules using psychological persuasion techniques. [The Verge]
And researchers found that AI hiring models consistently prefer resumes generated by themselves, potentially disadvantaging candidates who don’t use the same AI tool. [The Register]
Regulation
Colorado lawmakers delayed the state’s groundbreaking AI antidiscrimination law’s implementation from February to June 2026, hoping to refine its provisions in next year’s legislative session. [JD Supra]
Meta has strategically courted Trump, gaining support for its anti-regulation agenda in exchange for political donations and policy concessions. [Platformer]
Conversational AI
Eltropy has launched AI Voice Agents for community financial institutions, enabling natural, end-to-end conversational interactions that deflect 60% of agent calls. [EIN Presswire]
IntelePeer and Waterfield Tech have partnered to deliver an agentic AI platform that enables enterprises to modernize contact centers with minimal infrastructure changes. [Business Wire]
MyChatMap has developed a Cultural Personality AI platform that transforms enterprise content creation by analyzing workplace conversations across 70+ heritage combinations. [EIN Presswire]
PriceHubble has launched an AI Lead Calling Agent that automates lead qualification for real estate and mortgage professionals by making immediate, human-like phone calls. [EIN Presswire]
Salesforce has reduced its customer support workforce by nearly 50% by implementing its Agentforce AI platform, handling customer conversations with comparable satisfaction scores. [ITPro]
Taco Bell has scaled back its AI-driven drive-through technology after experiencing viral glitches and customer backlash. [eWeek]
Talkdesk has developed an advanced contact center integration for HubSpot, enabling seamless customer interaction and workflow management through AI-powered capabilities. [GlobeNewswire]
Be Real
Anthropic and other organizations are exploring ‘model welfare’, investigating whether AI models might be conscious and deserve moral considerations. [Wired]
DeepSeek emerged as a trusted AI health companion for a kidney transplant patient, offering comfort and medical advice in China’s overburdened healthcare system. [Rest of World]
Freelancers are finding new opportunities fixing AI-generated content that fails to meet professional standards across various creative and technical industries. [NBC News]
Hyodol has developed an AI eldercare robot that provides companionship and support to lonely seniors in South Korea. [Rest of World]
Moonlock reports a 148% surge in AI impersonation scams in 2025, using voice cloning and deepfake technology to convincingly mimic trusted individuals. [TechRadar]
Realbotix launched Ask Aria, a hyperrealistic AI chatbot enabling emotionally intelligent conversations in 25 languages on its website. [Business Wire]
Soul Machines launched Digital Workforce, an AI platform creating human-like digital workers with advanced interaction capabilities for enterprise businesses. [Business Wire]
Synthesia has developed an AI avatar technology that creates increasingly lifelike digital clones capable of mimicking human speakers with remarkable accuracy and expressiveness. [MIT Technology Review]
Toki AI launched an AI-powered avatar generator that transforms photos into lifelike, talking video avatars with natural expressions and personalized voices. [Speech Technology Magazine]
Wired and Business Insider published fabricated stories by a non-existent AI-generated journalist, exposing serious flaws in modern media fact-checking processes. [TechDirt]
Voice News
Aurelian has developed an AI voice assistant to handle non-emergency 911 calls, freeing human dispatchers to focus on critical emergencies. [TechRadar]
CallRail expanded its HubSpot integration to include Voice Assist, an AI-powered assistant that captures and qualifies leads 24/7 and syncs data seamlessly. [Business Wire]
Cerence AI and SiMa.ai have partnered to develop a low-power, intelligent voice interaction system for vehicles. [GlobeNewswire]
ElevenLabs launched ElevenReader, an AI audiobook platform offering authors a 60% royalty rate without exclusivity requirements. [Publishers Weekly]
Spoken has launched an AI-powered, multi-voice audiobook creation platform enabling authors to assign unique voices to characters. [Speech Technology Magazine]
Vocal Image, an Estonia-based startup, has developed an AI-powered app helping users improve communication skills through personalized voice coaching and interactive exercises. [TechCrunch]
Document AI
ABBYY and IBM have expanded their partnership to deliver an AI-driven, compliance-first KYC solution that transforms document processing and risk monitoring for financial institutions. [IBM]
AWS has developed an open-source GenAI IDP Accelerator that enables organizations to process millions of documents quickly and accurately using generative AI and serverless technologies on AWS. [Amazon]
Google Drive enhanced its Android document scanning experience with improved UI, cropping, rotating, and color adjustment tools, leaving iOS users without updates. [TechRadar]
Hyland has expanded its AWS partnership to develop AI-powered document processing solutions for regulated industries through enhanced cloud integration and co-development. [IDM]
Joomag has launched AI Article Extraction, transforming high-design PDFs into mobile-friendly, accessible, and SEO-ready digital articles. [EIN Presswire]
xSuite has released version 5.2.14 of its SAP-integrated Business Solutions, enhancing AI-powered invoice processing and intelligent document recognition across its workflow suite. [Business Wire]
Translation
DeepL launched its first AI agent, designed to automate repetitive business tasks across multiple languages for professionals in various sectors. [Tech.eu]
Google has upgraded its Translate tool with AI-powered live speech translation, enabling smoother, more natural conversations across languages. [Slator]
And Google has enhanced Circle to Search with continuous translation capabilities across apps and screens. [TechCrunch]
Newline Interactive has partnered with TranslateLive to integrate real-time translation technology into interactive panels, enabling multilingual communication across various professional environments. [PRWeb]
Welocalize and Duke University have unveiled LangMark, a large multilingual dataset designed to advance automatic post-editing research in MT. [MultiLingual]
Search
Google prevailed in avoiding Chrome divestiture but must share search data and limit exclusive contracts following a federal antitrust ruling against its search distribution practices. [Gizmodo]
Google has enhanced its AI Mode model to provide more concise and precise responses to complex STEM questions, targeting students returning to school. [Search Engine Land]
A survey by Higher Visibility reveals ChatGPT and AI tools rapidly gaining search market share, with Google's dominance declining from 73% to 66.9% between February and August. [Search Engine Land]
LLM Scout has launched a platform enabling businesses to track and enhance their brand visibility across AI search platforms like ChatGPT, Claude, and Google AI. [EIN Presswire]
Loxo launched an AI-powered Natural Language Search tool that enables recruiters to find candidates using plain language, simplifying the sourcing process. [PRWeb]
Perplexity offered students a free month of its $200 Pro AI browser Comet, with additional discounts to encourage continued use. [ZDNet]
Reddit is developing an AI-powered search feature that aims to transform the platform into a standalone search destination, leveraging its community’s expertise. [Fast Company]
Writing Assistance
Google has announced Android updates including an AI writing tool for Gboard. [ZDNet]
Mind Studio has developed an AI workflow automation platform that can generate complete, coherent 70,000-word novels from minimal input in just minutes, raising both technological excitement and ethical concerns about the future of literature. [DevX]
Wix introduced Email Assistant, an AI-powered tool that helps users draft, design, and refine marketing emails through conversational interaction. [TechRadar]
Health Tech
Medable launched Agent Studio, an AI platform for clinical development that aims to remove bottlenecks and accelerate drug trials through customizable, purpose-built agents. [Business Wire]
Phreesia launched VoiceAI, an AI-powered phone solution that helps healthcare organizations manage patient calls efficiently and improve patient experience. [Business Wire]
Primary Health Solutions has deployed SoundHound's AI-powered Denise platform to streamline patient appointment management and enhance customer experience. [Business Wire]
Rain Stella Technologies is developing voice-enabled operating room technology that aims to improve surgical documentation, efficiency, and patient safety through AI-powered assistants. [Gulf Business]
RAAPID launched an autonomous clinical AI platform powered by Neuro-Symbolic AI to dramatically improve risk adjustment accuracy and efficiency for healthcare organizations. [Business Wire]
Tolion Health is preparing to launch its AI-powered Tolion Brain Coach mobile app, aimed at improving brain health and longevity through personalized recommendations. [Business Wire]
ZyDoc has launched a multilingual translation service for medical notes, enabling global healthcare access through AI-powered, human-verified clinical documentation. [EIN Presswire]
Legal Tech
14OhFour has launched Lumina, an AI tool enabling law firms to quickly generate compelling event marketing content with minimal effort. [Artificial Lawyer]
8am MyCase has partnered with NetDocuments to enable seamless document management and access for law firms through an integrated platform. [Business Wire]
Eudia launched an AI-augmented law firm under Arizona’s Alternative Business Structure program, offering M&A due diligence and contract services with a proprietary intelligence system that learns from each engagement. [Artificial Lawyer]
Lake Merritt, an open-source AI evaluation system, empowers legal professionals to assess AI tools’ performance against specific business and legal priorities. [Artificial Lawyer]
LAW.co launched private, secure LLMs designed specifically for law firms to enable confidential, AI-assisted legal workflows while maintaining complete data control. [EIN Presswire]
LexisNexis and Thomson Reuters are investing heavily in legal tech, facing competition from startups while maintaining their dominant market position through strategic partnerships and technological innovation. [Business Insider]
LexisNexis has launched Protégé for PatentSight+, an AI-powered assistant that enables users to ask plain-language questions about patent data and receive structured, visual insights. [Artificial Lawyer]
LexisNexis reports that 61% of UK lawyers are using generative AI, but organizational cultures and heavy adoption remain slow, with only 11% using it extensively. [Legal IT Insider]
Mitratech partnered with iManage to embed document management capabilities directly into its CaseCloud platform, enabling seamless, secure legal workflow management. [Artificial Lawyer]
MyLegal acquired Leaflet's source code to create a unified legal tech platform that combines CLM, Gen AI, and DMS capabilities for corporate legal departments. [EIN Presswire]
NetDocuments has launched ndMAX Studio Apps, including a judicial analytics tool that creates judge profiles by analyzing past rulings using generative AI technology. [Artificial Lawyer]
Pangea researchers discovered a novel ‘LegalPwn’ attack that tricks LLMs into bypassing security guardrails by hiding malicious instructions within legal documents. [The Register]
Ed Tech
Atypical AI launched ExamJam SAT, an AI-powered platform providing personalized, adaptive SAT preparation through advanced learning science techniques. [GlobeNewswire]
GoMim AI launched an AI math solver that provides instant, step-by-step solutions to math problems across various educational levels and devices. [GlobeNewswire]
Broadcom Foundation has partnered with the Raspberry Pi Foundation to expand Experience AI, an AI literacy program aimed at educating teachers and students in Malaysia. [Business Wire]
Funding
ALIGNMT AI has raised US$6.5m to develop technology ensuring the safety and reliability of AI systems in healthcare. [Endpoints News]
Alpic has raised US$6m to develop a cloud platform for deploying, managing, and scaling MCP servers with advanced security and analytics. [Tech.eu]
Artificial Societies secured funding to develop AI simulation tools for testing ideas through digital personas. [Silicon Canals]
Exa, an AI search infrastructure startup, raised US$85m in Series B funding to enhance its search technology for AI agents. [Tech Funding News]
FriendliAI, an AI inference platform startup, has raised US$20m to help companies run AI models faster, cheaper, and more efficiently. [Crunchbase News]
Geniez AI secured US$6m in seed funding to connect LLMs and AI agents to mainframe data, enabling enterprise AI innovation. [Business Wire]
HappyRobot has raised US$44m to develop AI agents that automate operational tasks across logistics, freight, and customer support industries. [Tech.eu]
Hubert, an AI hiring platform, secured €2.5m in Seed funding to accelerate its autonomous interview screening technology for high-volume employers. [Tech.eu]
Intella secured US$12.5m in Series A funding led by Prosus Ventures to advance its Arabic speech intelligence technology. [FinSMEs]
Ketryx secured US$39m in Series B funding to help life sciences companies automate compliance and accelerate product development through its AI-powered platform. [Tech.eu]
Lambda, an AI infrastructure company, has hired banks for a potential IPO in the first half of 2026. [TechCrunch]
Matey, a legal AI startup, raised US$7.5m to expand CrimD, an AI platform that dramatically accelerates criminal defense case preparation and reduces associated costs. [Artificial Lawyer]
Mega secured US$2m in pre-seed funding to develop AI voice agents for streamlining order-to-cash management processes with a compliance-first approach. [Startups Magazine]
Mistral AI secured a €2B investment, valuing the French AI startup at €12B and establishing its prominence in European tech. [Bloomberg]
NewDays secured US$7m to develop an AI-powered therapy companion called Sunny, designed to support people with mild dementia through personalized conversations. [GeekWire]
Palabra AI has raised US$8.4m in pre-seed funding led by Alexis Ohanian to develop its advanced live speech translation technology across multiple languages. [Slator]
Pinecone, a top-funded AI database startup, has explored potential sale options amid the growing AI technology market. [The Information]
Predoc has raised US$30m to use AI for automating medical records management, helping doctors streamline administrative tasks and focus on patient care. [Fortune]
Propane, a Copenhagen-based AI customer intelligence startup, secured US$1.2m in pre-seed funding to develop its operating system. [Silicon Canals]
Recall.ai secured US$38m in Series B funding to expand its conversation data infrastructure API, enabling developers to build AI products that understand human interactions across various platforms. [Business Wire]
Sierra, a startup building custom AI agents for enterprise customer service, is close to securing a US$350m funding round valuing the company at US$10 billion. [Axios]
The Invigilator secured US$11-million in funding to expand its AI-powered educational technology platform internationally and enhance its assessment monitoring tools. [TechCentral]
Vouched secured US$17m in Series A funding to accelerate its Know Your Agent (KYA) platform, advancing trust and verification standards for AI agents and humans. [Business Wire]
You.com has raised US$100m in funding, valuing the AI startup at US$1.5 billion as it transitions from consumer search to broader AI services. [PYMNTS]
Acquisitions
AntWorks was acquired by GTT Data Solutions to expand its enterprise automation capabilities and global reach in the Intelligent Document Processing market. [Intelligent Document Processing Community]
Atlassian acquired Cycle, an AI-powered startup that helps companies transform customer feedback into actionable product development insights. [Tech.eu]
And Atlassian acquired The Browser Company for US$610m, aiming to develop an AI-powered browser optimized for knowledge work. [TechCrunch]
CoreWeave will acquire OpenPipe, a reinforcement learning platform that helps developers train more reliable and adaptable AI agents through advanced ML techniques. [Business Wire]
Databricks acquired Tecton to enhance its Agent Bricks offering by providing real-time contextual data for AI agents more efficiently. [InfoWorld]
Nvidia acquired Solver, a three-year-old AI coding startup that had raised US$8m in financing from investors. [The Information]
OpenAI acquired Statsig, a product testing startup, for US$1.1B in an all-stock deal, bringing its founder Vijaye Raji on as CTO of Applications. [TechCrunch]
PurpleLab acquired KAID Health to enhance its healthcare analytics platform by integrating AI-powered clinical insights and NLP technology. [Business Wire]
Speech Graphics acquired OC3 Entertainment, a facial animation technology developer known for FaceFX used in numerous AAA game franchises. [FinSMEs]
There’s More
ChatGPT consumes approximately 500 milliliters of water per short conversation, with usage varying widely based on location, cooling methods, and power generation. [Gizmodo]
Florida State University researchers discovered AI chatbots are increasingly influencing human language, with AI-specific words appearing more frequently in spontaneous conversations. [digitaltrends]
A study by LinkedIn revealed that a third of professionals feel embarrassed and nervous about their limited understanding of AI’s rapidly evolving landscape. [ZDNet]
Elon Musk has been aggressively promoting xAI's Grok chatbot by sharing sexualized anime-style images and interactions, drawing criticism for his seemingly obsessive focus on the AI’s erotic capabilities. [Rolling Stone]
Prominent AI researchers and tech leaders are increasingly describing AI in religious terms, framing it as a potential apocalyptic force that could either save or destroy humanity. [Broadband Breakfast]
Raft AI enabled US fighter pilots to receive real-time AI-assisted battle management directions during a groundbreaking military test. [Breitbart]
The US Special Operations Command sought to develop machine-learning-powered propaganda tools to influence foreign audiences and suppress dissenting arguments through automated information warfare systems. [The Intercept]
The Back Cover
Here’s Sierra Eagleson with a cover of Joni Mitchell’s A Case of You.
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!