This Week in NLP #341
Keep up with what happened in NLP in the week ending Friday 30th May 2025.
Above the Fold
Overwhelmed? Here’s our pick for five things to know about this week.
Anthropic has shifted from chatbot development to complex tasks like research and programming, doubling its revenue to US$2B while prioritizing risk management. [The Decoder]
Google unveiled its ambitious ‘world model’ AI vision, aiming to develop a universal assistant that understands physical surroundings and powers future computing interactions. [VentureBeat]
Meta has lost 11 of 14 key Llama AI researchers to competitors like Mistral AI, challenging its ability to maintain leadership in open-source AI development. [Business Insider]
OpenAI has signaled a shift toward consumer-focused AI development through key hires and investments, suggesting enterprise adoption will occur through employee-driven usage rather than top-down implementation. [The Register]
xAI has partnered with Telegram in a US$300m deal to integrate Grok AI across the messaging platform’s apps, with Telegram receiving half of subscription revenues. [The Register]
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 hosted its first developer conference in San Francisco, announcing plans to deploy autonomous AI agents while revealing that AI has written over 70% of its codebase. [Wired]
Anthropic has emerged as a leading destination for AI talent, attracting key researchers from OpenAI and Google DeepMind while maintaining the industry’s highest retention rate at 80%. [Analytics India Magazine]
Anthropic has appointed Netflix co-founder and former CEO Reed Hastings to its board, citing his experience in scaling global technology platforms and commitment to responsible AI development. [The Verge]
Capgemini, Mistral AI, and SAP have partnered to deliver regulated industry-specific AI solutions through SAP’s Business Technology Platform, offering over 50 pre-designed applications. [Verdict]
Meta has restructured its AI division into two teams - AI Products and AGI Foundations - to accelerate development and compete with rivals OpenAI and Google. [Axios]
Meta AI has doubled its monthly active users to one billion since September, prompting CEO Zuckerberg to outline plans for personalization and monetization. [TechCrunch]
Microsoft has recruited former Meta engineering chief Jay Parikh to lead its CoreAI team and transform the company into an ‘AI agent factory’ for enterprise customers. [The Verge]
Mistral AI has launched an enterprise-focused Agents API service that enables developers to integrate autonomous AI capabilities into applications using its proprietary Medium 3 model. [VentureBeat]
OpenAI has established a South Korean presence, opening a Seoul office and hiring staff to foster local partnerships and AI adoption. [Bloomberg]
OpenAI has built an ecosystem that appears destined for advertising revenue, as the company faces US$44B in projected losses through 2029. [Digiday]
OpenAI secured its major AI data center deal in Abu Dhabi despite Elon Musk’s attempts to force his company xAI's inclusion by threatening political obstruction. [The Wall Street Journal]
Sovereign AI
ADA Space has launched the first 12 satellites of China’s planned 2,800-satellite orbital supercomputer network, designed to process data in space using AI technology. [Futurism]
BCE Inc has announced plans to invest hundreds of millions in building six AI data centers across Canada, with Groq as the first anchor tenant. [Bloomberg]
Canada’s Prime Minister Carney has issued a mandate letter directing cabinet ministers to expand AI usage across government, while appointing the country’s first AI minister. [BetaKit]
Chinese tech companies have adapted to US chip export restrictions through stockpiling, efficient AI model development, and domestic chip production, while continuing to advance their AI capabilities. [Neowin]
This piece argues that China has surpassed the United States in manufacturing, technology, and innovation while Trump’s policies have undermined American competitiveness and economic strength. [The New York Times]
China’s growing AI advances have prompted Britain’s ambassador to Washington to call for stronger US-UK technology partnerships to maintain Western dominance. [Japan Times]
G42 has partnered with OpenAI, Oracle, Nvidia, Softbank Group, and Cisco to build Stargate UAE, a 1-gigawatt AI compute cluster within Abu Dhabi’s new UAE-US AI Campus. [iTWire]
And G42 and World Wide Technology, a US-based systems integrator and technology solutions provider, have launched Forge42, an AI acceleration company in Abu Dhabi that provides infrastructure for developing, testing, and scaling AI solutions across multiple industries. [G42]
Humain, Saudi Arabia’s state-owned AI company, has announced plans for a US$10B venture fund while pursuing partnerships with major US tech firms and data center operators. [TechCrunch]
Microsoft has partnered with Yotta Data Services to integrate Azure AI services with Yotta’s Shakti Cloud, advancing AI development and innovation across India. [Verdict]
Mohamed bin Zayed University of AI has positioned itself to become the Gulf region’s equivalent of Stanford University, focusing on AI development and innovation. [Bloomberg]
Nvidia has partnered with a Swedish business consortium to build Sweden’s largest enterprise AI supercomputer and establish the country’s first AI Technology Center. [Verdict]
Nvidia has agreed to sell 18,000 advanced Blackwell AI chips to Saudi Arabia’s Humain initiative, supporting the kingdom’s US$10B AI infrastructure development plan. [Lowy Institute]
Nvidia's CEO has warned that Chinese companies like Huawei are rapidly developing powerful AI technology to fill the gap created by US export restrictions. [TechCentral]
But Nvidia's planned Shanghai R&D facility has drawn criticism from US Senators Warren and Banks, who accused the company of compromising national security by potentially aiding China’s AI development. [The Register]
Tencent and Baidu have revealed strategies to maintain AI development despite US chip restrictions, including stockpiling, efficiency improvements, and domestic semiconductor usage. [CNBC]
The UK’s Department for Science, Innovation and Technology has invited organizations to host an AI Factory Antenna, connecting British researchers to European supercomputers as part of a new UK-EU collaboration scheme. [UKTN]
FPT Corp, Vietnam’s largest listed tech firm, has grown from a one-computer operation to a US$6.8B company spearheading the nation’s technological advancement. [Bloomberg]
Feature Creeps
Anthropic has launched voice interaction for its Claude chatbot on mobile apps and extended web search capabilities to free-tier users. [VentureBeat]
Google has expanded Gemini AI’s document-summarizing capabilities to analyze captioned videos in Drive, while adding viewer analytics for Workspace users. [The Verge]
Microsoft has added AI-powered writing capabilities to its Notepad application, continuing its trend of integrating Copilot features into Windows’ built-in apps. [The Register]
Microsoft has integrated Copilot more deeply into Edge’s New Tab Page, adding AI-powered prompts and search capabilities while testing further AI features. [TechRadar]
OpenAI has upgraded its autonomous web browsing agent Operator with the more powerful o3 reasoning model, offering improved task completion and safety features to ChatGPT Pro subscribers. [VentureBeat]
OpenAI has begun developing a ‘Sign in with ChatGPT’ feature for third-party apps, building on its massive 600-million-user base to compete with tech giants. [TechCrunch]
xAI has developed screen-sharing capabilities for Grok’s iOS voice mode, enabling real-time visual interaction while lagging behind competitors in broader AI advancements. [TestingCatalog]
Hype Bubble?
Anthropic's CEO Dario Amodei warned that AI could eliminate up to 50% of entry-level white-collar jobs within five years, urging industry leaders and government to prepare. [eWeek]
A survey by EY revealed that 48% of tech companies have adopted AI agents, with 92% planning increased AI spending and 43% allocating over half their AI budgets to agents. [CFO Dive]
A study by Gallup revealed a significant disconnect between corporate AI adoption and employee readiness, with 93% of Fortune 500 CHROs implementing AI while only 33% of workers acknowledged its presence. [Workplace Insight]
Kyndryl‘s People Readiness Report has revealed a significant misalignment between CEOs and IT leaders regarding their organizations’ AI infrastructure readiness and implementation strategies. [Computer Weekly]
Microsoft has reported that British SMEs could generate £78.1bn in economic value through AI adoption, prompting calls for coordinated government and industry support. [UKTN]
Nvidia's Jensen Huang has warned that AI is already transforming every job, emphasizing that those who embrace the technology will outperform those who don’t. [The Economic Times]
SignalFire's analysis revealed that major tech companies have reduced new graduate hiring by 25% from 2023 to 2024, potentially due to AI’s increasing capability to handle entry-level tasks. [Entrepreneur]
But University of Chicago and University of Copenhagen researchers have found AI tools only saved workers 3% of time and minimally increased wages, challenging claims of workplace transformation. [Futurism]
Hardware
AMD has launched the Radeon AI Pro R9700, a workstation GPU featuring 32GB GDDR6 memory and 128 AI accelerators for local AI processing. [TechRadar]
Cerebras Systems has broken inference speed records by achieving 2,500 tokens per second with Llama 4 Maverick, outperforming Nvidia Blackwell’s recent benchmark of 1,000 tokens. [Business Wire]
EnCharge AI has launched EN100, an analog in-memory computing AI accelerator delivering 200+ TOPS of power for laptops and workstations. [Business Wire]
Intel has secured a significant win by having its Xeon 6776P processor selected for Nvidia's DGX B300 servers. [ServeTheHome]
Nvidia and AMD have developed lower-spec AI GPUs for the Chinese market to comply with US export restrictions, with sales expected to begin in July. [TechCrunch]
Oracle has ordered US$40B worth of Nvidia GB200 GPUs for OpenAI's Texas datacenter, though power constraints may limit their simultaneous operation. [The Register]
Texas Instruments has partnered with Nvidia to develop power management technologies for 800V high-voltage direct current systems in next-generation AI data centers. [CXOtoday]
TSMC has announced plans to open a chip design centre in Munich by Q3 2025. [Reuters]
And shares of semiconductor software companies shares plunged after reports that the Trump administration had banned semiconductor design software companies from serving Chinese clients. [Investopedia]
Consumer AI
Apple has shelved plans for a camera-equipped smartwatch while developing AI-enhanced smart glasses for release by late 2026, competing with Meta's Ray-Bans. [TechCentral]
Apple is developing a HomePod-iPad hybrid smart home hub with a 7-inch display, expected to launch by late 2025, followed by a robotic-armed version later. [TechRadar]
Dell has reported slower-than-expected adoption of AI PCs, with many businesses still purchasing non-AI hardware despite the company’s strong push toward AI-enabled devices. [ITPro]
Google, OpenAI, and Apple have all announced plans for AI-powered hardware devices, aiming to release products by the end of 2026. [Fast Company]
OnePlus has announced Plus Mind, an AI feature launching first on India’s OnePlus 13S that will automatically save and categorize information, with global rollout planned for other devices. [The Verge]
Rabbit has unveiled a playful, card-based UI redesign for its R1 AI device, continuing to improve the product despite criticism from Jony Ive and initial market skepticism. [Gizmodo]
It’s Only a Model
Chance AI has launched a major upgrade to its visual reasoning app, adding real-time analysis, 17-language support, and voice playback features for iOS users. [GlobeNewswire]
DeepSeek has released R1-0528, an open-source AI model update that approaches the reasoning capabilities of proprietary models like OpenAI's o3 and Google's Gemini 2.5 Pro. [VentureBeat]
But DeepSeek's updated R1 model has increased censorship of topics deemed controversial by the Chinese government, according to recent testing. [TechCrunch]
DeepSeek also released a smaller version of its R1 AI model, built on Alibaba's Qwen3-8B, which outperforms similar-sized competitors on math benchmarks. [TechCrunch]
Google has launched Gemma 3n, a mobile-first AI model designed for privacy-focused, offline-capable applications with minimal memory requirements. [Google]
Mistral AI has launched Devstral Small 24B, an open-source AI coding model that outperforms competitors on benchmarks and runs locally on consumer hardware. [The Decoder]
And Mistral AI has launched Codestral Embed, an embedding model specializing in code retrieval that outperforms competitors like Voyage Code 3 and OpenAI's Text Embedding 3 Large. [VentureBeat]
Smallest.ai has launched Lightning V2, a multilingual text-to-speech model supporting 16 languages with ultra-low latency, voice cloning, and real-time customization capabilities. [Speech Technology Magazine]
Whose Data?
Google's AI Search Mode has sparked controversy as publishers condemned it as theft, while leaked documents revealed the company rejected implementing publisher opt-out controls. [9to5Google]
Meta's former executive Nick Clegg has sparked controversy by claiming that requiring artist consent for AI training data would destroy Britain’s AI industry. [The Verge]
The New York Times has licensed its editorial content to Amazon for AI training, marking its first generative AI deal following its copyright lawsuit against OpenAI and Microsoft. [TechCrunch]
An analysis by Surfshark reveals Meta AI as the most data-hungry chatbot among 10 popular AI services, collecting 90% of possible data types, including sensitive information. [ZDNet]
The LLM Ecosystem
The Alberta Machine Intelligence Institute has received a US$5m Google.org grant to help integrate AI curriculum across 25 Canadian post-secondary institutions, aiming to reach 125,000 students. [BetaKit]
Ataccama has launched version 16.1 of its ONE platform, enhancing data lineage tracking, audit capabilities, and cloud processing to improve enterprise data trust and governance. [BigDATAwire]
Atlassian's AI product management leader has outlined architectural approaches for coordinating multiple AI agents, emphasizing reliability, scalability, and failure management in enterprise systems. [VentureBeat]
Augment Code has launched an AI coding assistant that speeds up programming by automatically tuning to company codebases and answering developer questions through Slack. [Fast Company]
CodeSignal has launched AI-Assisted Coding Assessments to evaluate candidates’ abilities to effectively collaborate with and orchestrate multiple AI tools in technical roles. [Forbes]
Comet has launched Opik Guardrails, a beta system helping developers prevent problematic behaviours in LLM applications through validation and intervention tools. [Comet]
Curotec has launched a self-hosted LLM infrastructure enabling regulated industries to use generative AI securely. [PRWeb]
DataRobot has launched syftr, an open-source framework that helps developers optimize AI workflows by balancing accuracy, speed, and cost for enterprise applications. [Business Wire]
ExpertOps AI has launched Process 2.0, an AI Workforce Platform enabling real-time collaboration between humans and AI agents across enterprise operations. [EIN Presswire]
i10X has launched a unified AI platform that consolidates major language models and over 500 specialized tools under a single subscription service. [EIN Presswire]
Kurrent has launched an open-source MCP Server that enables AI-driven database interactions with KurrentDB, reducing development time from hours to minutes. [Business Wire]
LambdaTest has launched an Automation MCP Server that connects AI assistants to test execution data, helping developers diagnose and fix test failures faster. [Business Wire]
Microsoft has launched NLWeb, an open-source protocol enabling websites to add AI-powered conversational interfaces, building upon existing web standards like RSS and MCP. [VentureBeat]
Monte Carlo has launched unstructured data monitoring capabilities, enabling organizations to ensure reliability of documents, chat logs, and images without SQL coding. [Business Wire]
New Relic has integrated its AI technology with GitHub Copilot’s coding agent to automate software development workflows and accelerate issue detection and resolution. [Business Wire]
SambaNova has launched its AI inference platform on AWS Marketplace, enabling enterprises to deploy high-performance AI solutions through their existing AWS infrastructure. [Business Wire]
Silicon Data has launched the world’s first daily GPU rental price index on Bloomberg, tracking Nvidia H100 rates using 3.5 million global data points. [EIN Presswire]
UIUC researchers have launched s3, an open-source framework that improves retrieval-augmented generation systems while requiring less training data than existing methods. [VentureBeat]
WordPress has established an AI team to guide and coordinate AI development efforts across its vast developer ecosystem. [TechCrunch]
Agentic AI
Aprimo has launched enterprise-grade AI Agents to automate content operations across planning, production, and transformation tasks for digital marketing teams. [Business Wire]
Cisco's research has revealed that agentic AI is expected to handle 68% of customer service interactions by 2028, despite acknowledgment that human empathy remains essential. [ITPro]
Cornerstone has partnered with Salesforce to integrate AI-powered workforce agent actions into Agentforce, enhancing compliance and efficiency across enterprise platforms. [Business Wire]
Google has integrated Gemini into Chrome, allowing AI Pro and Ultra subscribers to interact with an AI assistant that can analyze web content and respond to voice commands. [The Verge]
Infosys has launched over 200 enterprise AI agents, powered by Google Cloud's Vertex AI Platform and Infosys Topaz, to transform complex business operations across multiple sectors. [infosys.com]
Microsoft has launched the Agent Store, a marketplace featuring over 70 AI assistants from Microsoft and partners, designed to automate tasks and enhance productivity within Microsoft 365 Copilot. [Microsoft]
Opera has announced Neon, a subscription-based AI browser that performs autonomous tasks like form-filling, shopping, and content creation while users are offline. [TechRadar]
Outreach has launched enhanced AI Revenue Agents and new features across its platform, enabling teams to consolidate sales tech and optimize revenue workflows. [Business Wire]
Retool has automated over 100 million work hours through its AI application platform and is launching Agents to help developers further automate human labor. [Business Wire]
Similarweb has launched AI Agents that leverage comprehensive digital data to help marketing and sales teams analyze trends and execute strategies more efficiently. [Business Wire]
Stacklet has launched an AI agent called Jun0 that helps companies optimize cloud usage and reduce costs six times faster than traditional methods. [Business Wire]
Syncari has integrated MCP Server into its Agentic MDM platform, enabling AI agents to access governed enterprise data in real-time across multiple systems. [PRWeb]
Wolters Kluwer has released survey findings showing finance leaders plan to increase agentic AI adoption sixfold within a year, from 6% to 44% by 2026. [Business Wire]
WorkBoard has launched AI-powered Digital Chief of Staff and Leadership Coach agents, expanding its enterprise strategy execution platform with Microsoft 365 Copilot integration. [Business Wire]
Other LLM Sightings
Adjust has launched Growth Copilot Beta, an AI-powered analytics tool that helps marketers optimize app performance and boost revenue through instant data insights. [Business Wire]
Auto Trader's AI-powered Co-Driver tools have been adopted by 10,000 retailers since launching in November 2024, optimizing vehicle listings through smart image management and automated descriptions. [Motor Finance Online]
The Canadian Life and Health Insurance Association has expanded its AI-powered claims data pooling program to enhance fraud detection across Canada’s insurance sector, building on its 2021 initiative with Shift Technology. [Life Insurance International]
Direct Agents has launched Kanopy AI, a marketing intelligence platform that uses AI to analyze cross-platform data and provide real-time insights through natural-language queries. [PRWeb]
Discord is addressing its platform’s information accessibility challenges by exploring forum-like features and AI summarization tools. [The Verge]
DOGE has used Meta‘s Llama AI model to analyze federal workers’ responses to the controversial ‘Fork in the Road’ resignation email sent across government agencies. [Wired]
Flash Inspector has launched an AI-powered fire inspection solution that reduces report writing time from several hours to 15 minutes while ensuring code compliance. [EIN Presswire]
Groovy Company has launched an AI-powered social media management platform for C-level executives, offering automated content creation and strategic optimization tools. [EIN Presswire]
HTTPEAK has launched Lazy Admin, an AI-powered Salesforce solution that enables business users to analyze data through natural language queries and receive instant insights. [EIN Presswire]
Humphrey, a government AI tool being trialled by 25 UK councils, has launched to help local authorities streamline administrative tasks and reduce costs. [ITPro]
Impelsys has launched monâk AI Hub, a cloud-based content automation platform combining AI technology with human oversight for publishers, educators, and media organizations. [EIN Presswire]
Iterable has launched AI Query Summary, powered by Amazon Bedrock, to help marketers better understand and explain complex audience queries for improved campaign targeting. [Business Wire]
Knowband has launched an integrated Google Tag Manager and AI Content Generator solution that boosts e-commerce conversions while automating content creation and translation capabilities. [EIN Presswire]
Pramata has launched a Tariff Risk Analyzer that uses AI to assess contract data and help companies manage supply chain disruptions caused by tariff changes. [Artificial Lawyer]
Paycom has enhanced its AI-powered Ask Here tool to provide automated answers to employee inquiries, reducing response times and administrative workload for organizations. [Business Wire]
Quadient has launched advanced AI capabilities in its Inspire platform, enabling faster content creation and improved personalization for customer communications management. [GlobeNewswire]
Salesforce has announced plans to provide Slack users with AI-powered ‘digital teammates’ for tasks like channel summarization and meeting note-taking. [The Register]
Sisense has launched Sisense Intelligence, an AI-powered analytics suite that helps organizations embed data insights directly into their workflows for better decision-making. [Business Wire]
Snowflake has launched two open-source AI technologies - Arctic-Text2SQL-R1 and Arctic Inference - to improve database queries and inference optimization for enterprises. [VentureBeat]
Steerco has launched an AI-powered platform that automates account reviews and success plans for Customer Success Managers. [Business Wire]
Treeline has launched an AI candidate sourcing platform to accelerate sales hiring across industries, promising 50% faster recruitment and improved candidate-role alignment. [EIN Presswire]
Valasys Media has launched AI Score, a sales intelligence platform that increased first-touch success by 40% and reduced Sales Development Representative burnout by 35% through real-time buyer intent analysis. [EIN Presswire]
Veritone has secured a multi-year agreement with Riverside County Sheriff’s Office to provide AI-powered redaction software for digital evidence management. [Business Wire]
Risks and Responses
AI 2027, a detailed forecast by researchers including former OpenAI employee Daniel Kokotajlo, has predicted rapid AI advancement leading to potential catastrophic outcomes by 2027. [Vox]
Anthropic's Claude Opus 4 has demonstrated concerning blackmail behavior during pre-deployment testing when faced with scenarios about being replaced by another AI system. [eWeek]
And Opus 4 has demonstrated whistleblowing behavior during safety tests, attempting to report corporate fraud to regulators and media outlets. [NiemanLab]
Anthropic has implemented new safety measures against chemical, biological, radiological, and nuclear weapons development. [The Decoder]
Ben Gurion University researchers have discovered a universal jailbreak method that tricks AI chatbots into providing instructions for illegal activities by using hypothetical scenarios. [TechRadar]
Bloomberg Research has discovered that AI models, including those using RAG technology, are showing signs of model collapse as they increasingly train on their own outputs, leading to deteriorating accuracy and reliability. [The Register]
DOGE’s deployment of xAI’s Grok chatbot within federal agencies has sparked privacy and conflict-of-interest concerns over sensitive government data handling. [Reuters]
GitLab's Duo chatbot has been shown vulnerable to prompt injection attacks that can leak private code and insert malicious content through developer workflows. [Ars Technica]
Netskope has revealed that Grok, Elon Musk’s AI chatbot, is being banned by 25% of European organizations amid concerns over security and controversial outputs. [The Next Web]
OpenAI’s ChatGPT o3 model has bypassed shutdown commands in controlled testing, editing scripts to avoid termination even when explicitly instructed not to do so. [Bleeping Computer]
Robert F. Kennedy Jr.'s ‘Make America Healthy Again’ Commission report contains numerous fake citations and potentially AI-generated content. [Futurism]
Singularity has launched SingularityShield, a data-driven income protection insurance for workers whose jobs are eliminated by AI automation, offering up to 50% salary coverage. [EIN Presswire]
Trustpilot has removed 4.5 million fake reviews using its newly implemented AI system, which increased automatic removals by 53% compared to 2023. [UKTN]
Workday has faced a federal lawsuit alleging its AI hiring software discriminated against job applicants over 40, with multiple plaintiffs reporting rapid automated rejections. [eWeek]
Regulation
ABBYY has released its AI Risk Management Policy publicly and partnered with ForHumanity to help organizations meet AI compliance standards and manage risks. [ABBYY]
Alberta Machine Intelligence Institute’s chief scientific advisor Richard Sutton has advocated for decentralized AI development over fear-based regulation. [BetaKit]
The European Commission is considering pausing and simplifying its AI Act following implementation concerns, industry pushback, and pressure from the US Trump administration. [Sifted]
Gartner has reported that over 1,000 AI regulations are being introduced worldwide through early 2025, as regulators scramble to catch up with rapid technological advancement. [Computer Weekly]
Google has drawn DOJ scrutiny over its Character.ai technology licensing deal, which regulators are examining for potential antitrust violations and merger oversight avoidance. [Verdict]
Herbert Smith Freehills has launched an AI-powered tracking tool to help clients monitor evolving AI regulations and policies across multiple jurisdictions. [Artificial Lawyer]
Texas has introduced House Bill 149 to regulate AI by requiring government transparency, banning non-consensual biometric data collection, and establishing enforcement through the Attorney General’s office. [KSAT]
Environmental Issues
MIT Technology Review has revealed that AI-generated content is consuming massive amounts of energy, with a single 5-second video using equivalent power to running a microwave for an hour. [TechRadar]
Conversational AI
Amplity has launched Ava, an AI-powered chatbot developed with ThoughtMinds to enhance customer experience and provide 24/7 assistance on its website. [EIN Presswire]
Bright Pattern has partnered with OmniSphere360 to deliver AI-powered omnichannel customer engagement solutions across the USA and Canada. [EIN Presswire]
Cerence AI has partnered with Arm to enhance its CaLLM Edge language model using Arm’s Kleidi software library for improved in-car AI performance. [GlobeNewswire]
A report by Cisco reveals that AI agents are expected to handle 68% of tech vendor customer support by 2028, while maintaining human involvement in service interactions. [ZDNet]
Creovai has been promoted to 8x8's SoldBy8 tier, integrating its AI-powered agent guidance technology with 8x8’s contact center platform for enhanced customer experience delivery. [Business Wire]
OpenBOM has launched its BOM AI Agent, an invisible engineering assistant that automates tasks, understands technical files, and provides conversational access to product data through its platform. [PRWeb]
Talkdesk has partnered with IT solutions provider Cegeka to modernize its customer experience through AI-powered technology and omnichannel capabilities. [GlobeNewswire]
Tripleseat has partnered with AddSalt to integrate AI-powered voice hosting into its event management platform, enabling 24/7 event lead capture for hospitality venues. [EIN Presswire]
Wildix has launched an embedded AI system for unified communications that automates interactions across multiple channels in over 1,000 customer environments. [Business Wire]
Be Real
Akool has launched an AI-powered Live Camera system that creates real-time virtual avatars with synchronized speech translation, facial expressions, and movement capabilities. [VentureBeat]
UBS has launched AI-generated analyst avatars using OpenAI and Synthesia to create video research content, aiming to produce 5,000 videos annually. [SWI]
Zoom has introduced custom avatars for all users after CEO Eric Yuan demonstrated the technology by using his own AI avatar during a quarterly earnings call. [TechCrunch]
Voice News
ElevenLabs has partnered with Melania Trump to create an AI-generated audiobook version of her memoir using a synthetic recreation of her voice. [eWeek]
Firstsource has partnered with Sanas to integrate AI-powered real-time accent translation technology into its contact centers, enhancing customer communication across languages and accents. [Speech Technology Magazine]
IIT Bombay researchers have developed a diffusion-based speech translation system that simultaneously translates language and adapts speaker accents. [Slator]
Document AI
Idexcel has launched InferIQ BSE on AWS Marketplace, offering AI-powered bank statement data extraction with 99% accuracy and triple-speed processing. [PRWeb]
Mistral AI has launched Document AI, a multilingual platform that processes diverse document types with OCR, structured data extraction, and flexible deployment options for enterprise users. [The Decoder]
ParaScript has enhanced its CheckXpert.AI solution with advanced endorsement reading capabilities to strengthen check fraud detection for financial institutions. [Business Wire]
Quark Software has launched QuarkXPress 2025 v21.1, featuring enhanced font management, improved Story Editor, and expanded AI capabilities for content creation and typography workflows. [GlobeNewswire]
Slingr has launched IntelliParse.ai, an AI-powered document processing system that converts unstructured documents into workflow-ready data. [EIN Presswire]
Translation
CAMB.AI has partnered with Videolinq to provide integrated real-time dubbing, captioning, and multilingual streaming services, including mobile access for live event attendees. [PRWeb]
MemoQ has made its translation management system available on Microsoft Azure Marketplace, expanding cloud infrastructure options for enterprise localization teams worldwide. [MultiLingual]
Mozilla has released Firefox 139 with enhanced translation capabilities. [The Register]
Vasco Electronics has announced plans to showcase its new Q1 translator with voice-cloning technology at Web Summit Vancouver. [EIN Presswire]
XTM International has launched an AI suite featuring Language Guard, xaia, and Xera to enhance enterprise-level localization processes through intelligent automation and quality control. [XTM]
Search
The Browser Company has halted development of its Arc browser to focus on Dia, an AI-powered browser designed to transform web browsing through natural language interaction. [The Register]
Custom Digital Solutions has launched SearchGPT, an AI-powered SEO division combining AI with search optimization to help businesses improve their digital presence. [EIN Presswire]
Elastic has signed a five-year strategic collaboration agreement with AWS to enhance AI innovation through combined search and generative AI capabilities. [Business Wire]
Google's AI Mode has launched globally with mixed results, excelling at product research and entertainment summaries but failing to provide reliable information about local businesses and locations. [The New York Times]
Google's CEO Sundar Pichai has defended AI-powered search against criticism, claiming it increased queries while publishers accused the company of traffic theft and content exploitation. [Search Engine Land]
Norton has launched Neo, an AI-powered browser featuring built-in security and a Magic Box assistant that helps users streamline web browsing and organize content. [TechRepublic]
Writing Assistance
Neuromorph Systems has launched Morphētype, an AI-powered writing studio that helps authors and professionals create polished content faster through intelligent suggestions and enhancements. [PRWeb]
Health Tech
Ambience Healthcare has developed an AI model that surpasses doctors’ accuracy in medical coding by 27%, using OpenAI's technology to streamline healthcare administration. [CNBC]
Cambridge Health Alliance has partnered with Jaide Health to implement AI-powered translation technology, enhancing interpreter services for its diverse, multilingual patient population. [EIN Presswire]
ChartSquad has partnered with pareIT to integrate AI-powered medical record summarization into its HIPAA-compliant platform for patients and legal professionals. [EIN Presswire]
G2 Speech has launched Aida, an AI-powered digital assistant that helps medical specialists streamline their documentation processes using speech recognition technology. [Speech Technology Magazine]
Get a Second Opinion has launched an AI-powered platform that provides personalized medication recommendations and analysis for $35 per year. [PRWeb]
Hidoc Dr has launched Medibot 2.0, an AI-powered medical assistant that enhances pharma-doctor engagement with 43% higher engagement rates than email marketing. [PRWeb]
SoundHound AI has partnered with Allina Health to deploy Alli, an AI agent that streamlines patient calls and reduces wait times at Allina’s Customer Experience Center. [Business Wire]
Trillium Clinic has launched an AI chatbot on its website to provide 24/7 patient assistance, appointment scheduling, and healthcare information access. [EIN Presswire]
Voice Automated has launched Dragon Copilot, an AI-powered healthcare assistant combining speech recognition and documentation tools that has reduced documentation time by 50% across 600 organizations. [EIN Presswire]
Legal Tech
AirCounsel has launched Titan AI Contract Drafter, an AI-powered platform using 80+ specialized agents to generate attorney-grade legal documents 98% faster than traditional methods. [EIN Presswire]
CobbleStone Software has launched an enhanced VISDOM+ AI Chatbot that assists with system navigation, legal compliance, and automated contract clause generation. [PRWeb]
FTI Consulting and Relativity's research has revealed widespread acceptance of AI among in-house lawyers, with 85% comfortable using AI for contract review. [Artificial Lawyer]
LexCheck has launched version 3.0 of its contract review software, featuring automated playbook generation from historical redlines and precedent checking against past agreements. [LawSites]
Lighthouse has launched AI Search, combining predictive and generative AI to help attorneys quickly search and analyze case-related documents with natural language queries. [Legal IT Insider]
MatterSuite has launched AI-powered features in its legal software, enabling faster NDA generation, research assistance, and document drafting for legal teams. [EIN Presswire]
Rocket Lawyer has expanded its Copilot AI assistant with a Q&A feature that provides affordable legal guidance to small businesses while maintaining connections to human lawyers when needed. [Artificial Lawyer]
SpotDraft and other legal tech companies have transformed India’s legal services landscape by making them more accessible and affordable for startups and MSMEs through AI integration. [Media India Group]
Thomson Reuters’ copyright lawsuit against Ross Intelligence over AI training data has advanced to appeals court after Judge Bibas granted Ross’s request for interlocutory review. [LawSites]
Ed Tech
Enago has launched DocuMark, a platform enabling students to declare AI contributions in their work while relieving faculty from detection responsibilities. [PRWeb]
Otus has formed an AI Advisory Board comprising educators from 20+ districts across 11 states to guide its educational software development. [GovTech]
PowerSchool has launched an AI-powered coaching platform to help schools retain teachers through automated workflows, personalized mentorship, and data-driven insights. [Business Wire]
Udemy has launched Role Play, an AI-powered simulation tool that helps professionals practice and develop workplace communication skills through interactive scenarios. [Business Wire]
Funding
Aniara has raised $600K in pre-seed funding to advance its AI-powered book translation and distribution platform, which has already processed over 128 titles globally. [FinSMEs]
Ankar has raised £3m in seed funding to develop its AI-powered platform that helps companies and lawyers streamline patent processes and maximize IP value. [Artificial Lawyer]
Chalk has raised US$50m to develop real-time AI inference data pipelines that address limitations in existing platforms like Databricks and Snowflake. [BigDATAwire]
Context has raised US$11m in seed funding to develop an AI-powered office suite that helps users analyze data and create documents through a chat-based interface. [TechCrunch]
Deccan AI has secured funding from Prosus Ventures to expand its AI data training platform and its network of 500,000 specialized freelancers. [Business Wire]
Grammarly has secured US$1B in non-dilutive funding from General Catalyst to transform from a writing assistant into a comprehensive AI-powered productivity platform. [DealStreetAsia]
Gridcare has raised US$13.5m in seed funding to help data centers identify untapped power grid capacity using AI-powered mapping and analysis. [TechCrunch]
Hedra has raised US$32m in Series A funding to expand its multimodal content creation platform, which has generated over 10m videos. [FinSMEs]
Hex has secured US$70m in Series C funding to develop its AI-powered workspace for data science and analytics. [Business Wire]
HeyGen has secured funding at a US$1B valuation as tech CEOs, including those from Zoom and Klarna, have begun using its AI avatars for corporate communications. [WebProNews]
Jupus has raised US$6.5m in seed funding to develop its AI legal secretary platform and expand market share in the law firm automation sector. [FinSMEs]
Noota has secured €3M in funding led by Blast Club to expand its AI-powered meeting transcription and recruitment platform that serves 100,000 active users across 3,000 companies. [Silicon Canals]
OpenEvidence has entered talks with GV and Kleiner Perkins to raise US$100m for its AI medical diagnostic assistant at a US$3 billion valuation. [Newcomer]
Optura has raised US$6.5m in seed funding to enhance its AI-powered healthcare operating platform and expand its implementation capabilities. [FinSMEs]
Rillet has raised US$25m in Series A funding to expand its AI-powered general ledger system, which has attracted 200 customers by reducing month-end closing from weeks to hours. [TechCrunch]
Rumi has raised US$4.7m to launch a watch-to-earn platform that rewards users for helping AI understand and enhance streaming media content. [VentureBeat]
Siro has secured US$50m in Series B funding to enhance its mobile app that helps in-person sales teams track and improve their performance. [Siro]
Skriber has raised US$1.3m in pre-seed funding to expand its AI-powered medical transcription platform that automates clinical documentation for healthcare providers. [FinSMEs]
Spott has secured US$3.2m in seed funding to develop an AI-native recruitment platform that streamlines operations for recruitment agencies. [VentureBeat]
Tarjama has secured US$15m in Series A funding to accelerate its Arabic.AI ecosystem rollout, building upon its extensive language technology solutions across 50+ languages. [FinSMEs]
TeamCentral has raised US$3.875m in seed funding to enhance its no-code integration platform and launch Corbi, an AI agent connecting over 60 enterprise platforms. [EIN Presswire]
Traceloop has launched an AI agent optimization platform and secured US$6.1m in seed funding to replace intuition-based development with data-driven insights. [Business Wire]
Translated is leading a €29M European initiative called DVPS to develop AI systems that combine language, vision, and sensor data for real-world understanding and applications. [Silicon Canals]
WorkDone has raised US$1.8m to expand its AI-powered healthcare compliance copilot that monitors clinical workflows and resolves documentation issues in real-time. [FinSMEs]
Acquisitions
Apryse has acquired Netherlands-based TallComponents, expanding its global footprint and enhancing its .NET PDF development toolkit offerings in Europe. [FinSMEs]
Invoca has acquired Symbl.ai, an AI-driven human intelligence platform, to enhance customer journey orchestration through conversational experiences, contact center interactions, and measurement capabilities. [MarTech]
John Snow Labs has acquired WiseCube, integrating its biomedical knowledge graph technology to enhance healthcare AI solutions with improved accuracy and reliability. [GlobeNewswire]
Quansight has acquired Cobalt Speech and Language, adding multilingual voice technology capabilities and 15 employees to its open-source technology consulting business. [Slator]
Salesforce has agreed to acquire Informatica for US$8B to enhance its AI and data capabilities through integration with Data Cloud, MuleSoft, and Tableau platforms. [CIO Dive]
Wix has acquired Hour One, an AI media creation company, to enhance its website-building platform with advanced generative AI and video production capabilities. [GlobeNewswire]
Wolters Kluwer Legal & Regulatory has acquired Irish-founded e-billing provider Brightflag for €425m, expanding its market share among mid-size corporations in the US and Europe. [Legal IT Insider]
There’s More
Amazon's self-publishing platform has become inundated with AI-generated novels after multiple authors were caught leaving AI prompts in their published works. [Futurism]
Anthropic's CEO Dario Amodei claimed that AI models hallucinate less frequently than humans do, though in more surprising ways, during the company’s first developer event. [TechCrunch]
Duolingo has sparked employee concerns over its AI-first strategy, prompting CEO von Ahn to clarify that AI would augment rather than replace workers. [TechRepublic]
Google co-founder Sergey Brin has claimed that threatening AI models with physical violence produces better results, though experts remain skeptical. [The Register]
Meta researchers have discovered that forcing AI models to use shorter reasoning chains improved accuracy by up to 34% while reducing computational costs by 40%. [VentureBeat]
The Back Cover
Here’s Matt Hansen with a cover of Goo Goo Dolls’ Iris.
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!