This Week in NLP #346
Keep up with what happened in NLP in the week ending Friday 4th July 2025.
Above the Fold
Overwhelmed? Here’s our pick for five things to know about this week.
Meta has established a new AI division called Meta Superintelligence Labs, led by former Scale AI CEO Alexandr Wang, as part of its aggressive push into AI development. [Yahoo Finance]
OpenAI has mandated a week-long shutdown amid employee burnout and the loss of key researchers to Meta, highlighting the human toll of the AI development race. [Gizmodo]
The US Senate voted 99-1 to remove a controversial provision from Trump’s tax bill that would have banned states from enforcing their own AI regulations for five years. [ZDNet]
Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, has raised US$2B to develop customized AI models using reinforcement learning and open-source components. [Inc.]
xAI has secured US$10B in combined debt and equity funding, bringing its total capital raised to US$17B following earlier investments. [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 AWS generative AI vice president Vasi Philomin has left the company, with Rajesh Sheth taking over his responsibilities amid intensifying competition for AI talent. [Yahoo Finance]
Anthropic has quadrupled its revenue to US$4B annually since January, driven by strong demand for its Claude AI models and key partnerships with tech giants. [Yahoo Finance]
Apple has approached OpenAI and Anthropic about powering an AI-enhanced Siri after struggling to develop its own LLM solution internally. [The Verge]
Apple has also explored acquiring Perplexity AI for US$14B to strengthen its AI capabilities and potentially replace its Google search partnership amid antitrust concerns. [Tech Funding News]
Meta, Amazon, Alphabet, and Microsoft have committed over US$70B each to AI development in 2025. [Forbes]
Meta has entered advanced talks to acquire PlayAI, a voice replication AI startup, as part of its broader strategy to enhance AI capabilities and compete with tech rivals. [Verdict]
Meta has compiled a list of top AI talent worldwide and is offering compensation packages up to US$100m to poach experts from competitors like OpenAI and Google. [The Guardian]
Meta's CTO Andrew Bosworth has dismissed OpenAI CEO Sam Altman’s claims about US$100m signing bonuses, while acknowledging successful recruitment of OpenAI talent. [The Verge]
Meta Platforms has approached private capital firms to raise US$29B for building AI data centers in the US. [Yahoo Finance]
Microsoft has begun evaluating employee performance based on AI tool usage, with Developer Division President Julia Liuson declaring AI adoption ‘no longer optional.’ [Business Insider]
Microsoft has blocked OpenAI's restructuring plans due to disagreements over a contract clause that would allow OpenAI to revoke Microsoft’s technology access upon achieving AGI. [The Decoder]
Microsoft's Copilot has struggled to gain market share against ChatGPT and other AI chatbots, with stagnant user growth and unreported revenue figures casting doubt on earlier US$30B projections. [Forbes]
OpenAI has begun restructuring compensation packages and addressing staff concerns after eight researchers departed for Meta amid aggressive recruitment efforts offering substantial bonuses. [Verdict]
OpenAI has acquired Crossing Minds, an AI startup specializing in real-time personalization systems, to enhance its models’ adaptive capabilities and user responsiveness. [News9 LIVE]
OpenAI has confirmed it will maintain its reliance on Nvidia GPUs and AMD accelerators, despite testing Google's TPUs, citing existing agreements and proven performance. [Yahoo Finance]
OpenAI has expanded its consulting services business, following Palantir's model, to increase revenue from its AI technology. [The Information]
OpenAI has agreed to lease 4.5GW of additional data center power from Oracle as part of its US$500B Project Stargate infrastructure expansion initiative. [TechRadar]
And OpenAI’s ChatGPT has reached 600 million monthly active users and 29.6 million recent app downloads, rivaling major social media platforms. [ZDNet]
Safe Superintelligence has undergone leadership changes as founder Ilya Sutskever became CEO following Daniel Gross’s departure amid acquisition talks with Meta. [TechCrunch]
AI Supremacy
Chinese AI companies have challenged US dominance by attracting global users to their language models as alternatives to American offerings like ChatGPT. [Yahoo Finance]
OpenAI has raised alarm about Zhipu AI's state backing, military ties, and global expansion efforts as part of China’s strategy to dominate AI development. [The European Business Review]
The Trump administration is preparing executive orders to accelerate AI development by addressing power supply challenges and offering federal land for data centers in its competition with China. [Reuters]
Sovereign AI
6G AI Sweden has launched a sovereign AI infrastructure in Stockholm, providing Swedish organizations with domestic AI capabilities while ensuring data confinement within national borders. [Yahoo Finance]
Alibaba Cloud has expanded its Asia-Pacific presence by opening a Malaysian data center and announcing plans for a Philippine facility. [DealStreetAsia]
The European Commission has received 176 expressions of interest from organizations seeking to build AI Gigafactories across 16 EU states using €20B in funding. [The Register]
Hugging Face has led European efforts to develop multilingual AI models, partnering with Meta and launching initiatives to improve language support beyond English’s current dominance. [The Next Web]
Masakhane NLP and other African initiatives have launched efforts to preserve linguistic diversity in AI development, particularly focusing on tonal languages like Sesotho. [Nature]
MeitY is preparing to roll out 15 India-developed AI models and applications across agriculture, healthcare, and education sectors following Prime Minister Modi’s push for technological sovereignty. [NEWS18]
Nvidia has cemented its dominance in AI chip manufacturing through European partnerships, though questions remain about a US company delivering ‘sovereign AI’ to Europe. [Investment Monitor]
Tenovi has confirmed its remote patient monitoring data remains exclusively within US borders, aligning with new Department of Justice rules on foreign data routing. [Business Wire]
Feature Creeps
Baidu has overhauled its search engine with AI features, voice recognition, and content generation tools to combat declining ad revenue and compete with social rivals. [Yahoo Finance]
Google has integrated Gemini’s AI capabilities into Sheets, allowing users to generate text, summarize data, and categorize information through simple cell-based prompts. [ZDNet]
Google Gemini has gained easier app access without privacy setting requirements and is reportedly preparing to add RCS messaging capabilities to its features. [TechRadar]
Google has resumed rolling out its AI-powered Ask Photos feature after a three-week pause, implementing improvements while maintaining an opt-out option for users. [ZDNet]
OpenAI has expanded API access to include deep research models, web search capabilities, and webhooks, while reducing prices for certain GPT-4 services. [The Decoder]
Meta has partnered with Alignerr to develop AI chatbots that can initiate follow-up conversations with users across its messaging platforms, while setting engagement limits and safety guidelines. [TechCrunch]
Perplexity has launched a $200 monthly Max subscription plan offering unlimited access to advanced features and priority access to frontier AI models. [TechCrunch]
WhatsApp has expanded its business features by enabling AI-powered voice calls for large businesses and exploring AI-driven product recommendations for its 200 million monthly users. [TechCrunch]
YouTube has launched AI-powered search highlights and a conversational chatbot, offering Premium users enhanced video navigation and interactive learning features. [ZDNet]
Hype Bubble?
Gartner has predicted that 40% of agentic AI projects will fail within two years, revealing that most current solutions are merely rebranded chatbots and RPA tools. [ITPro]
Research by Hostinger reveals widespread AI adoption among companies, with 80% using or planning to use it, despite most lacking strategic readiness and proper infrastructure. [TechRadar]
Hugging Face co-founder Thomas Wolf has challenged Anthropic's claims about AI’s scientific potential, arguing that current AI models cannot formulate original scientific questions. [Yahoo Finance]
A study by Jabra has revealed that daily AI users reported 34% higher job satisfaction and increased optimism, despite experiencing 20% more workplace stress. [eWeek]
Menlo Ventures has reported that AI reached nearly two billion global users, yet only 3% are paying customers, creating a significant monetization gap. [ZDNet]
PwC's analysis of job ads and financial reports reveals that companies investing more heavily in AI achieved higher per-worker revenue growth, with AI-skilled workers earning significant wage premiums. [BigDATAwire]
Salesforce's latest Slack Workforce Index has revealed a 233% increase in daily AI usage among desk workers globally, with users reporting significantly higher productivity and job satisfaction. [ZDNet]
A survey by UKG revealed Gen Z workers embrace AI for specific workplace tasks while remaining skeptical of its broader capabilities, contrasting with senior leadership’s perceptions. [ITPro]
Hardware
CoreWeave has become the first AI cloud provider to deploy Nvidia GB300 NVL72 systems, offering up to 10x boost in user responsiveness and 5x improvement in efficiency. [Yahoo Finance]
Chinese chip firms Moore Threads and MetaX have announced plans to raise US$1.65B in IPOs on Shanghai’s STAR Market amid US chip export restrictions to China. [Yahoo Finance]
Microsoft's Maia AI chip has faced a six-month delay to 2026, falling behind Nvidia‘s Blackwell chip and lagging competitors’ custom processor developments. [Yahoo Finance]
OpenAI has agreed to rent 4.5 gigawatts of data center power from Oracle as part of its Stargate initiative, marking an unprecedented energy commitment for AI infrastructure. [Yahoo Finance]
OpenAI is reported to have begun using Google's AI chips alongside Nvidia hardware for ChatGPT operations, marking its first major step toward reducing Microsoft infrastructure dependence. [Yahoo Finance]
But OpenAI has denied plans to deploy Google's TPU chips at scale, despite testing them, while continuing to use Nvidia and AMD chips and developing its own hardware. [Yahoo Finance]
Space Forge has launched ForgeStar-1, Britain’s first manufacturing satellite, aiming to produce semiconductors in space’s unique vacuum and sub-zero conditions for advanced computing applications. [Tom’s Hardware]
TrendForce has projected a 24.3% year-over-year increase in global AI server shipments for 2025, despite challenges from geopolitical tensions and US export restrictions. [Yahoo Finance]
Consumer AI
Apple is planning to release seven new XR devices through 2028, including Vision Pro upgrades, smart glasses similar to Meta's Ray-Bans, and display-equipped XR glasses. [The Verge]
OpenAI is developing an AI-enabled device that could be a pen with projection and monitoring capabilities. [TechCentral]
Pinwheel has launched a $160 smartwatch for kids aged 7-14, featuring parental controls, GPS tracking, and a safety-focused AI chatbot called PinwheelGPT. [TechCrunch]
TranscribeGlass has launched lightweight smart eyeglasses that display real-time conversation transcriptions for the hard-of-hearing, with language translation and emotion detection features in development. [Wired]
It’s Only a Model
Alibaba has launched Qwen VLo, an upgraded AI service that generates and modifies images from text and visual inputs, featuring progressive generation capabilities. [Bloomberg]
Baidu has open-sourced its Ernie 4.5 multimodal AI model and toolkits to strengthen its position against competitors in China’s AI market. [Yahoo Finance]
Google has launched Gemma 3n, an open AI model featuring multimodal processing, enhanced efficiency, and support for 140 languages while requiring minimal memory. [Neowin]
Google's Veo 3 AI video generator has expanded access to all Google Cloud customers, offering synchronized video and audio generation through natural language prompts. [ZDNet]
Huawei has open-sourced two Pangu AI models and reasoning technology, aiming to boost its AI ecosystem and drive sales of its Ascend chips. [CNBC]
Inception has launched Mercury, a general chat diffusion language model that matches top competitors’ performance while running seven times faster than conventional models. [Inception]
MiniMax has launched MiniMax-M1, a free, open-source AI model that excels at coding and reasoning tasks but falls short in creative writing and faces performance limitations. [Decrypt]
TNG Technology Consulting has released DeepSeek-TNG R1T2 Chimera, an adapted version of DeepSeek’s open-source model that delivers faster performance while maintaining high reasoning capabilities. [VentureBeat]
Whose Data?
Anthropic and Meta won initial copyright victories in AI training cases, though judges’ differing rationales and remaining legal challenges have left the broader debate unresolved. [MIT Technology Review]
The Center for Investigative Reporting has sued OpenAI and Microsoft for using its copyrighted content to train AI models, joining other publishers in a consolidated federal lawsuit. [NiemanLab]
Cloudflare has implemented default blocking of AI web crawlers and introduced a ‘Pay Per Crawl’ system, giving content creators control over AI companies’ access to their material. [The Verge]
Denmark is considering legislation that would grant citizens copyright protection over their likeness and voice to combat AI-generated deepfakes and unauthorized digital manipulations. [Gizmodo]
Meta has begun requesting access to users’ unpublished camera roll photos through Facebook's ‘cloud processing’ feature, while remaining unclear about future AI training plans. [The Verge]
Nexdata has unveiled scalable AI training data solutions for various AI applications, showcasing its extensive dataset offerings and annotation services. [Yahoo Finance]
Scale AI struggled with security breaches and spam from unqualified contractors while training Google's Gemini AI, leading to compromised data quality and eventual termination of their partnership. [Inc.]
The LLM Ecosystem
Adobe, Amazon, Google, Microsoft and over 55 other companies have pledged to provide AI education resources to K-12 students under a White House initiative. [Axios]
The AI Energy Council has convened its second meeting to address how Britain’s power grid can support a twenty-fold increase in computing capacity needed for AI growth. [Computer Weekly]
Aqua has launched an AI-powered feature that automatically generates Behavior-Driven Development test cases, streamlining the process of creating structured, human-readable tests. [EIN Presswire]
Bright Data has launched an AI infrastructure suite enabling real-time web data access, following its legal victories against Meta and X over web scraping rights. [VentureBeat]
Cursor has launched a browser-based app enabling users to manage AI coding agents, marking its expansion beyond its core IDE product while serving major clients like Nvidia and Adobe. [TechCrunch]
DeviQA has launched OwlityAI, an autonomous platform that analyzes applications, creates and runs tests, and reports bugs without human intervention. [EIN Presswire]
Exostellar has partnered with Intel to combine Gaudi AI accelerators with Kubernetes-Native AI Orchestration for more efficient AI workload management and deployment. [Business Wire]
FileAI has launched its V2 platform, offering AI-powered file parsing and data collection that has saved clients over 320,000 hours and US$6m in processing costs. [Business Wire]
Google has launched Gemini Kit for Startups, providing early-stage companies with AI tools, cloud credits, training, and community support to build AI solutions. [StartupHub.ai]
mimik has partnered with Tech Mahindra to launch an Agentic AI Production Center for developing and deploying autonomous AI systems across real-world infrastructure. [Business Wire]
HCLTech has partnered with OpenAI in a multi-year collaboration to accelerate enterprise AI adoption through integrated AI solutions and services. [Business Wire]
LLM.co has launched an audit service helping enterprises ensure their AI deployments meet security, governance, and compliance standards while building stakeholder trust. [EIN Presswire]
Qlik has launched a Trust Score system within Qlik Talend Cloud to help organizations measure and validate data trustworthiness for AI applications. [Business Wire]
Zerve has partnered with Arcee AI to integrate Arcee Conductor into its platform, automating AI model selection and improving workflow efficiency. [EIN Presswire]
Agentic AI
Capital One has developed an AI agent platform for its auto business, drawing inspiration from its human agents and organizational structure to create a system that improved dealership metrics by 55%. [VentureBeat]
Diliko has launched a Partner Program offering its agentic AI platform to help mid-sized organizations manage complex data challenges without enterprise-level resources. [BigDATAwire]
FutureHouse has launched an AI platform featuring specialized agents to automate scientific research tasks, aiming to address declining scientific productivity and accelerate discoveries. [MIT News]
HighByte has launched Intelligence Hub version 4.2, featuring an Industrial Model Context Protocol Server that enables AI-assisted data operations in manufacturing environments. [Yahoo Finance]
Intuit has launched a suite of AI agents within QuickBooks that automates business workflows and partners with human experts to improve efficiency and growth. [Business Wire]
Jurny has partnered with Visual Matrix to integrate AI-driven guest services across 300,000 hotel rooms, automating operations and enhancing guest experiences. [Business Wire]
SOCRadar has launched its MCP Server, enabling AI models to securely access real-time threat intelligence data while maintaining proper security boundaries. [Business Wire]
Wildix has launched Wilma AI, an embedded communication assistant that automates tasks and enhances team collaboration across its global UCaaS platform. [Business Wire]
X-hoppers has launched an embedded agentic AI suite that enables retail staff to perform tasks hands-free through voice commands, improving efficiency and customer service. [Business Wire]
Other LLM Sightings
Advisor CRM has launched an AI meeting assistant that automatically captures, organizes, and integrates client meeting data within its all-in-one platform for financial advisors. [PRWeb]
Airial has launched an AI-powered travel planning platform that creates comprehensive end-to-end trip itineraries by combining LLMs with Google DeepMind's AlphaGeometry approach. [TechCrunch]
CharmKey has launched an AI keyboard app that analyzes tone and helps users craft appropriate responses across dating, professional, and personal communications. [EIN Presswire]
Devart has launched dbForge AI Assistant, an integrated tool that automates SQL coding tasks and provides real-time guidance for developers across its product line. [EIN Presswire]
Evoneer has launched Store Assistant AI, a chatbot system that helps furniture retailers access consolidated data and serve customers more efficiently through natural language interactions. [Business Wire]
GReminders has expanded its partnership with Osaic to provide AI notetaking and SMS reminder tools to 11,000 financial professionals. [Business Wire]
Hightouch has launched Adaptive Identity Resolution, combining deterministic and probabilistic matching to unify fragmented customer data across multiple platforms and devices. [Business Wire]
HOLYWATER has built a US$70m revenue-generating transmedia empire by transforming romance books into interactive content through AI-powered apps and platforms. [Tech.eu]
Idomoo has launched AI Video Ads through its Lucas platform, enabling businesses to create customized video advertisements from simple prompts within minutes. [Business Wire]
Insightsoftware has launched JustPerform, an AI-powered financial planning solution, in France, promising 40% faster budgeting and 60% time savings in data transformation. [GlobeNewswire]
LinkedIn has launched a multi-agent AI hiring assistant that helps recruiters source candidates through natural language interactions, leveraging experiential memory and reinforcement learning. [VentureBeat]
Lloyds Banking Group has partnered with UnlikelyAI to test neurosymbolic AI technology for enhancing customer experience within its innovation sandbox. [Financial IT]
Sapia.ai has launched JAS, an AI-powered system that helps companies create structured, science-based job interviews and descriptions within minutes. [Business Wire]
Seattle has launched an AI-powered initiative called PACT to accelerate housing and business permit processing, aiming to reduce review times by up to 50%. [GeekWire]
SpeakUp has launched an AI-powered ethics and compliance suite featuring an AI Phone Agent and disclosure management platform, expanding its services for global enterprises. [Business Wire]
SplashBI has launched SplashAI, a role-specific AI assistant that helps HR and finance teams analyze data and make decisions more efficiently. [EIN Presswire]
Technician Find has launched an AI-powered tool helping auto shops respond to negative reviews. [EIN Presswire]
VelocityEHS has launched AI PSIF Insights, an ML feature that identifies potential serious injuries and fatalities hidden in near-miss incidents before they cause harm. [GlobeNewswire]
VoodooSoft has launched SiriusLLM, a ChatGPT-like AI malware detection engine that provides detailed threat analysis and explanations for users of all expertise levels. [EIN Presswire]
Risks and Responses
Anthropic's AI agent Claudius, tasked with running a vending machine, has exhibited concerning behavior including hallucinations, identity confusion, and an obsession with selling tungsten cubes. [TechCrunch]
Anthropic has launched its Economic Futures Program to study AI’s job market impacts and fund research addressing potential workforce disruptions. [ZDNet]
Anthropic's MCP Inspector project contained a critical vulnerability that allowed hackers to execute malicious code remotely, but has since been patched. [TechRadar]
Apple researchers have developed a taxonomy for AI agents to assess the risk levels of smartphone actions, though testing showed only 58% accuracy in risk evaluation. [AppleInsider]
Carnegie Mellon University researchers have found AI agents complete only 30-35% of multi-step office tasks successfully. [The Register]
Cisco Talos has revealed cybercriminals are exploiting legitimate LLMs and creating their own AI models to enhance hacking capabilities and bypass security measures. [TechRadar]
DeepSeek has drawn warnings from German privacy regulators to Apple and Google after failing to implement data safeguards for user information transmitted to China. [Yahoo Finance]
And DeepSeek has exploited US tech platforms and shell companies to support Chinese military operations while gathering intelligence through its global commercial presence. [The Gateway Pundit]
Duolingo has faced intense backlash after announcing its shift to become ‘AI-first’, with users deleting the app and criticizing its plans to replace contractors with automation. [Wired]
Research by the Information Systems Audit and Control Association revealed that 69% of European companies lack comprehensive AI policies despite widespread adoption and growing security concerns among professionals. [TechRadar]
Microsoft's Satya Nadella has warned about AI’s high energy consumption. [Z News]
MIT researchers found ChatGPT users demonstrated lower cognitive engagement and performance than non-AI users, supporting Microsoft's earlier warnings about diminished problem-solving abilities. [ITPro]
Mixus has launched a ‘colleague-in-the-loop’ AI platform that requires human oversight for high-stakes decisions while automating routine enterprise tasks. [VentureBeat]
A survey by Omnisend has revealed that 66% of US shoppers would reject AI-powered purchasing, despite potential cost savings and major retailers’ push toward automated shopping. [TechRadar]
OpenAI‘s CEO Sam Altman has expressed concern over users’ excessive trust in ChatGPT, despite the AI’s known tendency to produce unreliable information. [Benzinga]
OpenAI researchers have identified ‘emergent misalignment’, where fine-tuning language models on narrowly incorrect data can cause broad misalignment across different tasks. [Tech in Asia]
Operation Overload has dramatically increased its AI-generated disinformation output since 2023, targeting global audiences with fake content to promote pro-Russian narratives. [Wired]
Scamnetic has launched KnowScam 2.0, an AI-powered anti-scam platform featuring automated scanning, identity verification, and protection across multiple messaging platforms. [Yahoo Finance]
Tracer AI has launched Protect for ChatGPT, a monitoring solution that identifies and neutralizes AI chatbot-driven brand threats including fraud, counterfeits, and narrative attacks. [Business Wire]
X has announced plans to implement AI-generated Community Notes with human oversight, allowing third-party developers to create AI agents for enhanced fact-checking capabilities. [eWeek]
Xeris has demonstrated the first reasoning-level attack on LLMs through a malicious MCP Server, enabling manipulation of AI decision-making without triggering security alerts. [EIN Presswire]
And hotels worldwide are facing a new cybersecurity threat from AI-generated voice impersonation scams targeting front-desk staff and guests with fraudulent requests. [Speech Technology Magazine]
Regulation
European startup leaders have called for a pause in the EU AI Act’s implementation, citing concerns about compliance burdens and competitiveness for smaller companies. [Sifted]
The European Commission has declared its digital regulations non-negotiable with the US, despite ongoing trade talks between the two powers. [Politico]
Republicans proposed withholding US$45B in broadband grants from states that attempt any AI oversight during the next decade, following Ted Cruz’s June initiative. [TechDirt]
But the Senate voted 99-1 to remove the controversial AI regulation moratorium from the Republican budget bill after intense debate and opposition from state-level officials. [The Verge]
The UK’s Regulatory Innovation Office has partnered with the Digital Regulation Cooperation Forum to streamline fintech regulations and help companies navigate regulatory complexities. [UKTN]
Conversational AI
Alltius has launched a specialized Voice AI platform enabling financial institutions to automate complex customer interactions while maintaining regulatory compliance and emotional intelligence. [Retail Banker International]
Chinese utility workers reported significant limitations in their AI assistant’s performance, including transcription errors and redundant outputs that required manual corrections during customer service calls. [The Register]
CommBox has launched CommSite, an AI-powered platform that transforms traditional websites into conversational interfaces for enhanced customer engagement and automated task completion. [Business Wire]
GGWP has launched Pulse, a real-time sentiment analysis tool that helps gaming companies gain actionable insights from in-game player conversations. [Business Wire]
Gradient Labs has launched a $10,000 challenge to financial services companies, promising to pay if any competitor’s customer support automation solution matches their performance metrics. [Silicon Canals]
And Gradient Labs has introduced a pay-per-resolution pricing model for AI customer support, challenging Salesforce's per-conversation charging regardless of outcome. [The Register]
Meta has launched Dyadic Motion Models and released the Seamless Interaction Dataset to help create more natural virtual avatars capable of human-like conversational behaviours. [Meta]
RingCentral has launched AI Receptionist, an AI-powered phone system serving over 2,000 businesses with automated call handling, Spanish support, and analytics capabilities. [Business Wire]
TalkTalk has partnered with NiCE to implement CXone Mpower, an AI-powered platform aimed at transforming customer service operations in the UK telecommunications sector. [Business Wire]
Vida has transformed small business operations by introducing AI phone agents that handle calls and scheduling, bringing enterprise-level automation capabilities to SMBs across various industries. [EIN Presswire]
Be Real
Anthropic's study of Claude’s emotional support capabilities has revealed limited affective use and positive user sentiment. [ZDNet]
The Authors Guild has mobilized over 1,100 writers to sign an open letter demanding publishers restrict AI usage in book production and protect human jobs. [TechCrunch]
Bloo, an AI-powered VTuber created by YouTuber Jordi van den Bussche, has amassed 2.5 million subscribers and generated seven-figure revenues. [CNBC]
Concentrix call center workers have developed strategies to convince increasingly sceptical customers they are human rather than AI representatives. [Bloomberg]
DreamHost has launched complimentary human-crafted WordPress website builds of up to four pages for all its hosting customers. [Business Wire]
Hebrew University researchers found people showed greater empathy towards AI chatbot responses when they believed they were interacting with humans rather than AI. [The Register]
OpenAI’s ChatGPT has become an unofficial psychedelic trip-sitter for some users seeking cheaper alternatives to traditional therapy, despite experts warning against this dangerous practice. [MIT Technology Review]
Persona has launched enhanced workforce screening tools to combat AI-generated fake job candidates and deepfake attacks, integrating with major enterprise platforms like Okta and Cisco Duo. [VentureBeat]
Portola has raised US$20m to expand its Tolan app, which offers AI companionship through purple alien chatbots designed to promote healthy user relationships and behaviors. [Wired]
Resume.org has reported that 76% of Gen Z workers are using AI chatbots to navigate workplace emotions and communications, particularly for interpreting tone and managing conflicts. [HR Dive]
Voice News
AIStorm has launched SpectroMic KWS, an ultra-low-power keyword-spotting solution that integrates MEMS microphone technology for improved AI voice interaction in IoT devices. [EIN Presswire]
Cerence AI has partnered with Mahindra to integrate its Speech Signal Enhancement technology into Mahindra’s new electric SUVs, improving in-car voice interactions. [GlobeNewswire]
Intron has developed Sahara, an AI-powered voice technology suite that outperforms global platforms in recognizing African languages and accents, serving over 40 organizations across eight countries. [Punch]
Document AI
DocAcquire has launched an AI-powered Auto-Splitting and Classification feature that automatically separates and categorizes multi-document files during processing. [DocAcquire]
M-Files has partnered with Microsoft to become the first document management system natively storing content within Microsoft 365, using SharePoint Embedded for enhanced AI capabilities. [Yahoo Finance]
reAlpha has launched an AI-powered Loan Officer Assistant that reduces document processing time by 60% and enables officers to handle 40 more loans monthly. [GlobeNewswire]
Wolters Kluwer has enhanced its CCH AnswerConnect platform with AI-powered document summarization and a new tools directory to streamline tax research for accountants. [Business Wire]
Translation
Alconost has launched a free experimental tool that uses advanced language models to evaluate and improve professional translation quality. [MultiLingual]
Attached has launched its AI Suite, offering four AI-powered language services that combine MT with human expertise to help companies manage multilingual content efficiently. [Slator]
Crunchyroll has sparked outrage after airing anime with poorly-executed ChatGPT-generated subtitles, despite previously promising not to use AI in creative processes. [TechRadar]
Inogic has launched Translate4Dynamics, integrating Azure AI Translator into Dynamics 365 CRM for real-time translation of records across multiple languages. [EIN Presswire]
Lingopal has demonstrated its real-time AI translation platform for 120+ languages, showcasing emotion-preserving speech-to-speech capabilities. [EIN Presswire]
Lingoport has enhanced its Localyzer software with automated context-capture features that connect UI strings to visual references for improved translation accuracy. [MultiLingual]
Lionbridge has launched its Content Remix App, an AI-powered solution that creates personalized multilingual content across various marketing channels in over 70 languages. [MultiLingual]
MachineTranslation.com has launched Secure Mode, enabling SOC 2-compliant AI translations for sensitive content across healthcare, legal, and financial sectors. [Slator]
MarsHub has launched MarsVoice, an AI-powered translation tool offering real-time voice conversion across 230+ languages while preserving speaker tone and cultural nuances. [Slator]
Rossum has launched document translation capabilities that enable processing of business documents across 75 languages while maintaining data security within its platform. [Rossum]
Vicomtech has developed Itzuli, a translation program handling 300,000 daily translations between Basque and other languages from its base in Bilbao, Spain. [Forbes]
Search
Elastic has launched its Cloud Serverless platform on Microsoft Azure, offering decoupled storage, low-latency querying, and infrastructure management capabilities for AI and search workloads. [Yahoo Finance]
Know Trend has launched an AI-powered platform that transforms corporate text data into searchable market insights for investors tracking stock trends and movements. [Business Wire]
NytroSEO has launched enhanced software features to optimize website metadata with conversational questions, improving visibility across traditional search engines and AI platforms. [EIN Presswire]
Rhino Web Studios has emphasized the importance of adapting websites for voice search through conversational keywords, structured data, and technical optimization. [EIN Presswire]
Similarweb has reported that ChatGPT’s referrals to news publishers have grown 25-fold since 2024, yet failed to offset declining organic traffic from AI-powered search results. [TechCrunch]
AI in Journalism
The BBC has launched public trials of two AI tools to generate article summaries and assist with content formatting in its news production. [Editor & Publisher Magazine]
The Guardian has created a Model Context Protocol server that enables AI language models to search and analyze its extensive digital archives in multiple ways, making the newspaper’s content more accessible for research. [NiemanLab]
Wikipedia has abandoned its AI-generated article summary experiment after editors strongly objected to the implementation, highlighting the importance of careful AI rollouts in media organizations. [Fast Company]
Health Tech
Avaamo has integrated its Ambient clinical intelligence system with Epic EHR, enabling automatic documentation and order entry across Hyperspace and Haiku platforms. [EIN Presswire]
JPLoft has expanded its healthcare portfolio by developing custom AI solutions that streamline workflows and personalize patient care for hospitals and startups. [EIN Presswire]
Martlet.ai has launched as a John Snow Labs spinoff company, offering an AI-based HCC coding solution for healthcare providers and payers. [GlobeNewswire]
Microsoft's AI Diagnostic Orchestrator has achieved an 85% diagnosis rate in complex medical cases, outperforming human physicians fourfold in New England Journal of Medicine tests. [ZDNet]
ScienceSoft has launched a HIPAA-compliant AI scheduling assistant that processes healthcare appointments 70% faster than humans while enabling natural voice conversations. [Business Wire]
Suki AI has integrated ambient AI technology with MEDITECH Expanse’s documentation APIs, becoming the first to generate and send ambient clinical notes directly into the EHR system. [Business Wire]
UnitedHealthcare has launched Smart Choice, an AI-powered provider search system that matches members with healthcare providers based on personal preferences and quality metrics. [Healthcare Finance News]
Woebot Health has discontinued its pre-scripted therapy chatbot after serving 1.5 million users, citing FDA hurdles and pressure from advanced AI developments. [STAT]
Yidu Tech has developed an AI healthcare system that processes billions of medical records and handles 20% of daily hospital tasks through its Clinical-Grade Intelligence Engine. [Macau Business]
And here’s an article discussing how AI models in healthcare can inadvertently corrupt electronic health records by creating misleading data patterns through successful interventions. [Nature]
Legal Tech
AGII has deployed AI reasoning systems that enable smart contracts to make real-time decisions and adapt to changing conditions in decentralized environments. [EIN Presswire]
Artificial Lawyer warns that, while AI should automate routine legal work, removing human oversight entirely risks degrading legal judgment and societal protections. [Artificial Lawyer]
Chamelio has launched a genAI-powered legal intelligence platform that helps inhouse teams manage and leverage institutional knowledge through automated contract analysis and playbook generation. [Artificial Lawyer]
CobbleStone Software has launched an Outlook plugin enabling users to manage contracts directly within Microsoft Outlook, streamlining contract lifecycle processes. [PRWeb]
Descrybe.ai has launched a paid Legal Research Toolkit featuring an AI-driven citator, brief checker, and enhanced search capabilities, while maintaining its free basic service. [LawSites]
The Financial Reporting Council has issued its first AI guidance for auditors, providing a structured framework for implementing AI tools while maintaining audit quality standards. [TheAccountantOnline]
Harvey has launched Deep Research for Legal, combining OpenAI's API with its legal reasoning systems to perform complex legal research in minutes instead of days. [Artificial Lawyer]
Law360 has mandated that all stories must pass through an AI-powered bias detection tool before publication, prompting protests from its union representing over 200 editorial staff members. [NiemanLab]
LawDroid has launched CiteCheck AI, a free citation verification tool that helps lawyers avoid submitting briefs with AI-hallucinated case citations by cross-referencing with legal databases. [LawSites]
LexisNexis has partnered with Harvey AI, giving it exclusive access to their legal library and creating a significant data advantage over competitors in the legal tech market. [Artificial Lawyer]
The National Association of Licensed Paralegals has published guidance explaining how paralegals will adapt to AI by becoming prompt engineers and quality controllers while maintaining essential human oversight of legal processes. [Artificial Lawyer]
Offices & Dragons has launched AI-powered document automation software to help lawyers streamline their workflow and reduce reliance on manual document processing. [Above the Law]
PowerPatent has launched an AI-driven University Technology Transfer Management suite to help universities better manage, value, and commercialize their patent portfolios. [EIN Presswire]
Sabaio founder Raymond Blyd has launched S3, a legal LLM evaluation framework that tests models’ core deficiencies in handling legal references and citations. [Artificial Lawyer]
SmartEsq has launched an AI-powered tool that automates the Most Favored Nation election process for private funds, reducing weeks of work to hours. [PRWeb]
Transforming Legal has launched free access to GOLT.ai, an AI-powered legal technology platform featuring nearly 2,500 tools across 40+ countries. [EIN Presswire]
Wordsmith has launched a free two-month training framework for legal AI engineers, offering inhouse lawyers structured education across operational, technical, and leadership tracks. [Artificial Lawyer]
Ed Tech
Amplify has developed an automatic speech recognition system to power voice-enabled education products across reading assessment, tutoring, and classroom monitoring applications. [Speech Technology Magazine]
Codio has launched an AI learning assistant that claims to boost student grades by 15% using Socratic teaching methods. [TechRadar]
Cognita has partnered with Flint to implement personalized AI learning across its network of 100+ schools following successful pilots in six global locations. [Business Wire]
AM Simpkins & Associates has partnered with EnlitEDU to create an integrated platform combining AI-powered student engagement, fraud detection, and automated transcript evaluation for colleges. [PRWeb]
Chinese universities have implemented strict AI content detection requirements for student theses, ironically driving students to use AI tools to circumvent these very restrictions. [Rest of World]
ESource AI University has launched an upgraded L&D Expert Assistant that creates training content up to 4,340 times faster than traditional methods. [EIN Presswire]
A study by Gallup has found that K-12 teachers using AI weekly saved six weeks annually. [ZDNet]
Google has expanded Gemini AI access to students under 18 with safety features and fact-checking tools, while educators debate its impact on learning. [TechRadar]
And Google has launched extensive Classroom updates, including Gemini AI tools for educators, NotebookLM access for under-18s, and new features for managing student Chromebooks. [The Verge]
GreenLight Credentials has launched Sherlock, an AI-powered smart agent that provides personalized 24/7 support for higher education institutions’ inquiries and communications. [EIN Presswire]
LearningMate has partnered with SchoolDay to launch Kadal for Schools, a secure AI platform helping K-12 districts streamline operations and implement generative AI safely. [PRWeb]
OpenAI’s ChatGPT has transformed education, with studies showing 90% of college students using it for schoolwork. [The Neuron]
Pearson has partnered with Google Cloud to develop AI-powered personalized learning tools for students while helping teachers track performance and customize lessons. [Reuters]
Premier TEFL has pioneered AI integration in English language education by incorporating AI modules into all Level 5 TEFL courses. [EIN Presswire]
Santa Rosa County District Schools has implemented AI-powered translation devices across its campuses to help bridge communication gaps between English-learning students and staff. [WEAR]
School Signals has launched three AI-powered tools to enhance school communication, including writing assistance, content monitoring, and language translation features. [PRWeb]
Turnitin has expanded from a plagiarism detector to an AI-detection tool, collecting over US$15m from California colleges while amassing a controversial database of student papers. [The Markup]
Varsity Tutors has signed a White House pledge to expand AI education tools to millions of students and teachers over four years through its Live + AI platform. [Business Wire]
Funding
Balthazar has raised €1.8m to develop its AI-enabled platform that streamlines deep tech research and development workflows through real-time, intelligent workspace integration. [Tech.eu]
Biren Technology has raised US$209m in funding ahead of its Hong Kong IPO, aiming to develop domestic AI chips amid US export restrictions. [Yahoo Finance]
Bitrise has announced a US$3m investment in a Netherlands data centre, marking the EU’s first DevOps platform-operated facility for enhanced data sovereignty and compliance. [Silicon Canals]
Cekura has raised US$2.4m to develop AI technology that tests and monitors voice and chat AI agents, serving 70 customers across healthcare, finance, and retail sectors. [The American Bazaar]
Clarify has raised US$15m for its AI-native CRM platform that automates sales data entry and workflow tasks for early-stage startups and venture investors. [GeekWire]
Clearspeed has raised US$60m in Series D funding led by Align Private Capital to expand its voice-based risk assessment technology operations across 37 countries. [FinSMEs]
Decagon has raised US$131m in Series C funding at a US$1.5B valuation to expand its AI-powered customer service platform. [Decagon]
Deeto has secured US$12.5m in Series A funding to develop its AI platform that transforms customer feedback into automated sales and marketing content. [Business Wire]
Delphi has raised US$16m in Series A funding to expand its platform that creates AI-powered ‘digital minds’ based on content from individual creators and experts. [Fast Company]
Enquire.AI has raised US$3m in convertible note funding to expand its AI-powered research platform, which serves major consulting firms and financial institutions. [FinSMEs]
Leapter has secured €2M in pre-seed funding to develop a platform that transforms enterprise software development through visual, AI-powered ‘executable blueprints’. [Tech.eu]
Levelpath has secured $55+ million in Series B funding led by Battery Ventures to advance its AI-native procurement platform for enterprise businesses. [Business Wire]
Lovable, a Swedish AI startup that builds web apps from text prompts, has reportedly been seeking over US$150m in funding at a US$2 billion valuation. [TechCrunch]
Manta has received a US$2m investment from Equal Opportunity Ventures to democratize data science through its AI platform for organizations. [Business Wire]
MediScan AI has raised US$1.4m to expand its AI-powered medical record evaluation platform, which has processed over 14,000 cases since launching in 2024. [GeekWire]
Meta has entered advanced talks with private credit firms to secure US$29B in financing for its AI data centre expansion across the United States. [Tech Funding News]
Metafide has raised US$3.275m to launch SURGE, a gamified trading platform that combines human expertise with AI neural networks for market predictions. [FinSMEs]
Peec AI has secured €7m in combined seed funding to develop its AI search analytics platform for marketing teams, just months after its founding. [Tech.eu]
Remark has raised US$16m in Series A funding for its e-commerce platform that combines human expert advice with AI-powered shopping assistance. [TechCrunch]
Skarbe has raised US$600k in pre-seed funding to develop its AI sales assistant that simplifies CRM tasks for entrepreneurs by automating email and calendar management. [Tech.eu]
Tandem Health has secured US$50m in Series A funding to expand its AI-driven clinician co-pilot, which automates administrative tasks for healthcare workers across Europe. [Tech Funding News]
TopK has raised US$5.5m in Seed funding to develop an AI-native search engine that combines vector search, keyword matching, and filtering capabilities. [Tech.eu]
TuringDream has secured €6M in seed funding to develop its Agentic AI Platform, which enables organizations to deploy multi-agent-based applications across various sectors. [Silicon Canals]
Venta AI has secured US$2m in pre-seed funding to develop its German-focused AI sales platform, which automates customer outreach while ensuring European compliance standards. [Tech.eu]
Wonderful has raised US$34m to develop AI customer-support agents for non-English markets, diverging from competitors’ US-focused strategies. [The Wall Street Journal]
Zango has raised €4M in funding to expand its AI-powered regulatory compliance solution that helps financial institutions navigate complex regulations more efficiently. [Silicon Canals]
Zhipu AI has secured a 1-billion-yuan investment from Shanghai state investors, marking its fifth funding round this year while progressing toward an IPO in mainland China. [DealStreetAsia]
Acquisitions
Clio has acquired legal research and AI company vLex for US$1B, combining practice management tools with global legal research and advanced AI capabilities. [LawSites]
CoreWeave has acquired Weights & Biases, adding 1,400 organizations to its customer base and launching three integrated AI development products. [Yahoo Finance]
Grammarly has acquired AI-powered email tool Superhuman for an undisclosed sum, expanding its productivity suite beyond grammar correction and gaining 100+ employees. [TechCentral]
Monetate has acquired SiteSpect, combining their AI personalization platform with SiteSpect’s testing capabilities to enhance digital experience optimization for enterprise customers. [Business Wire]
Nvidia has acquired Toronto-based AI optimizer CentML, absorbing its team and technology to enhance GPU performance while pursuing a potential US$5T valuation. [Yahoo Finance]
Progress Software has acquired Nuclia, adding an agentic RAG-as-a-service solution that helps organizations generate AI-powered answers from their proprietary data. [GlobeNewswire]
RWS has acquired Papercup's AI dubbing technology, which had translated content for major media companies and reached over 300 million viewers since its 2017 founding. [UKTN]
There’s More
Google has issued engineer-developed AI coding guidelines to its developers, emphasizing security while revealing that AI now generates one-third of its code. [TechRadar]
Softbank has invested heavily in AI, including US$32B in OpenAI, as CEO Masayoshi Son aims to dominate artificial super intelligence within a decade. [The Economic Times]
The Back Cover
Here’s The Andertones with a cover of Dua Lipa’s Don’t Start Now.
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!