This Week in NLP #347
Keep up with what happened in NLP in the week ending Friday 11th July 2025.
Above the Fold
Overwhelmed? Here’s our pick for five things to know about this week.
Mistral AI, the French AI startup behind Le Chat chatbot, is seeking up to US$1B in equity funding from investors including Abu Dhabi’s MGX fund. [TechCrunch]
Nvidia has become the first publicly traded company to reach a US$4T market valuation, driven by surging demand for its AI-powering chips. [Associated Press]
OpenAI has developed an AI-powered browser that will compete with Chrome and keep users within ChatGPT’s ecosystem rather than linking to external websites. [TechCrunch]
And Perplexity has launched Comet, a Chromium-based browser featuring built-in AI search capabilities and natural language interactions, challenging Google's browser dominance. [The Register]
xAI has launched Grok 4 and a $300 monthly SuperGrok Heavy subscription, while facing controversy over its AI’s antisemitic posts on X. [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 has considered expanding its investment in Anthropic beyond its existing US$8B commitment to maintain its position ahead of Google. [Yahoo Finance]
Anthropic‘s EMEA head Guillaume Princen has denied targeting rival labs’ talent while overseeing a regional expansion that will double the team to 200 employees. [Tech.eu]
Meta has assembled a new superintelligence division led by Alexandr Wang, poaching researchers from OpenAI, Google DeepMind, and other companies. [eWeek]
Meta offered Apple's AI chief Ruoming Pang a US$200m package to join its superintelligence division, while Apple declined to counter the offer. [Investopedia]
Meta has poached nine OpenAI employees with large signing bonuses for its superintelligence division, prompting criticism from Sam Altman about the recruitment tactics. [eWeek]
Meta has established a £12m research facility in Cambridge to develop spatial audio technology for its wearable AI products. [UKTN]
OpenAI has launched a six-month residency program offering $105,000 to train non-AI specialists, countering Meta's aggressive poaching of established AI talent. [Yahoo Finance]
And OpenAI has quintupled its stock-based compensation to US$4.4B in response to Meta's poaching of key researchers. [WinBuzzer]
OpenAI has recruited four prominent engineers from Tesla, xAI, and Meta to strengthen its scaling team, which manages critical infrastructure for AI model training. [Wired]
OpenAI’s ChatGPT has doubled its user base and increased news-related queries by 212% while driving significant traffic to partner news sites, as Google's news referrals declined. [The Decoder]
Perplexity has been identified by Wedbush analyst Dan Ives as Apple's best chance to catch up in the AI race. [Fortune]
Replit has partnered with Microsoft to offer its app-building platform through Azure Marketplace while integrating with various Microsoft cloud services. [TechCrunch]
Safe Superintelligence has appointed OpenAI co-founder Ilya Sutskever as CEO, following Daniel Gross’s departure to Meta's new superintelligence division. [eWeek]
Tesla is preparing to integrate xAI's controversial Grok chatbot into its vehicles next week, offering multiple AI personalities despite recent issues with inappropriate responses. [TechCrunch]
xAI has purchased an overseas power plant for relocation to the US to support a new data center requiring up to 2 gigawatts to power one million AI GPUs. [Tom’s Hardware]
AI Supremacy
Chinese tech firms have planned installations of 115,000 Nvidia AI chips across western China, despite US export controls prohibiting such acquisitions without licenses. [Yahoo Finance]
Nvidia's CEO Jensen Huang has planned a strategic visit to Beijing amid US-China tech tensions and export restrictions threatening billions in Chinese market revenue. [Yahoo Finance]
And Nvidia is developing a modified AI chip for the Chinese market that complies with US export restrictions, aiming for a September release. [TechCrunch]
Meanwhile, the US government has lifted export restrictions on semiconductor design software to China, allowing Synopsys, Cadence, and Siemens to resume full services to Chinese customers. [Verdict]
The US government has also proposed new export controls on AI chips to Malaysia and Thailand to prevent their potential diversion to China. [The Decoder]
Sovereign AI
Centraleyes has launched an EU AI Act compliance framework within its platform as Europe invests €20B in AI infrastructure and implements strict regulations. [EIN Presswire]
Civo's CEO criticized the UK government’s £2bn AI investment strategy for continuing to rely on US cloud giants rather than building domestic technological sovereignty. [CityAM]
The European Commission has received 76 proposals from major tech companies to build AI Gigafactories across 16 EU countries, aiming to deploy three million GPUs. [AI News]
Eviden has completed a major upgrade to Brazil’s Santos Dumont supercomputer, boosting its AI capabilities and energy efficiency as part of the national AI plan. [HPCwire]
Google Cloud has launched a UK startup accelerator program and new data residency controls, offering AI firms technical support, cloud credits, and UK-based data processing capabilities. [ITPro]
And Google Cloud has launched UK-based data processing for Gemini 2.5 Flash AI, though support remains global and US authorities may still access data. [The Register]
India’s government has launched a US$1.25B AI initiative to develop indigenous language models and AI infrastructure, responding to China’s DeepSeek breakthrough in early 2025. [MIT Technology Review]
Ooredoo has launched Qatar’s first sovereign AI cloud platform, powered by Nvidia GPUs and operated by local data center company Syntys, supporting national digital priorities. [RC Wireless News]
Feature Creeps
Baidu has launched MuseSteamer, a business-focused AI video generator, alongside major search engine updates to compete with rival AI platforms. [Reuters]
Coinbase has partnered with Perplexity AI to integrate real-time crypto market data into Perplexity’s Comet browser, enabling AI-powered market analysis. [The Street]
Google’s Firebase Studio has introduced Agent Mode with Gemini CLI integration, enabling AI-powered code generation with varying levels of autonomy. [The Register]
Google has integrated Gemini’s customizable AI chatbots (Gems) into Workspace apps’ side panels, enabling direct access to specialized AI assistance without switching applications. [The Verge]
And Google has announced five new Gemini AI features for Android devices, including enhanced foldable phone support, improved Circle to Search, and smartwatch integration. [TechRadar]
Google has expanded Gemini to Wear OS watches and enhanced Circle to Search with AI Mode integration, while adding gaming query support and improved AI Overviews. [TechCrunch]
Google has also expanded AI Mode in India to all users without requiring Search Labs sign-in, just two weeks after its initial launch in the region. [Search Engine Land]
Microsoft has rolled out AI-powered search capabilities in Copilot for Windows 10/11, enabling natural language queries and in-file content extraction. [TechRadar]
Zoom has expanded its AI Companion’s capabilities to integrate with 16 third-party apps and cross-platform meeting services. [ZDNet]
Hype Bubble?
Carnegie Mellon researchers have found that AI agents, including Google's Gemini 2.5 Pro, failed most real-world office tasks despite massive venture capital investment. [Futurism]
Companies have been paying human experts substantial fees to fix mistakes made by AI systems they had adopted to cut costs. [Futurism]
Confluent's AI entrepreneur Sean Falconer has argued against the hype of open-world AI agents, advocating instead for closed-world, event-driven enterprise solutions. [VentureBeat]
Ford CEO Jim Farley has predicted that AI would eliminate approximately half of all white-collar jobs in the United States. [TechRadar]
Microsoft has reported significant AI-driven productivity gains across its operations, including US$500m in call center savings, while simultaneously reducing its workforce by 15,000 employees. [Yahoo Finance]
Research by Qlik revealed that only 11% of UK businesses have achieved tangible gains from their AI initiatives, with most companies still stuck in experimental phases. [TechRadar]
But Responsive's research has shown that nearly two-thirds of UK and EU B2B revenue leaders achieved ROI within their first year of AI adoption. [ITPro]
A survey by ThinkAutomation revealed that tech leaders expect AI to transform 47% of existing roles by 2030. [EIN Presswire]
Hardware
Alphabet's Tensor Processing Units have emerged as a potential challenger to Nvidia's GPU dominance in AI infrastructure, particularly for inference workloads. [The Motley Fool]
Amazon has built Project Rainier, a massive AI supercomputing cluster using hundreds of thousands of Trainium2 accelerators, for Anthropic across multiple US datacenters. [The Register]
Arm has expanded its data center customer base to 70,000 since 2021, while doubling its applications to 9 million and growing its developer base to 22 million. [Yahoo Finance]
CoreWeave has launched Nvidia's RTX PRO 6000 Blackwell Server Edition at scale, becoming the first cloud provider to deploy this powerful AI-focused GPU technology. [Yahoo Finance]
Huawei has begun offering its older Ascend 910B AI chips to Middle Eastern and Southeast Asian markets, though production limitations and uncertain US policy responses complicate expansion plans. [Yahoo Finance]
IBM has launched Power11 data center chips and servers, offering enhanced power efficiency, AI integration capabilities, and robust security features for business operations. [Yahoo Finance]
Premio has launched its LLM-1U-RPL edge server, featuring Intel processors and Nvidia GPU support for on-premises AI processing and real-time GenAI workloads. [EIN Presswire]
Super Micro Computer has announced plans to expand its AI server manufacturing across Europe, building on its existing Netherlands operations to meet growing regional demand. [Yahoo Finance]
Consumer AI
Meta has invested US$3.5B in EssilorLuxottica, acquiring a 3% stake in the Ray-Ban maker to strengthen its position in smart glasses. [Bloomberg]
Samsung has launched natural language routines and multiple feature updates for SmartThings, including improved device controls and expanded Calm Onboarding capabilities. [The Verge]
It’s Only a Model
Alibaba Group has led Chinese tech companies in releasing open-source AI models through ModelScope, which now serves 16 million developers across 36 countries. [China Daily]
Cerebras Systems has launched Qwen3-235B, a frontier AI model offering 131K context support and delivering code generation at 30x the speed of alternatives. [Business Wire]
ETH Zürich has trained two open-source language models on Switzerland’s Alps supercomputer, promising full transparency with source code and weights under Apache 2.0 licensing. [The Register]
Huawei has open-sourced its Pangu AI model series, including a 72B-parameter MoE model and an upcoming 7B-parameter version, with Ascend chip optimization. [The Decoder]
Huawei has denied accusations from HonestAGI that its Pangu Pro AI model copied elements from Alibaba's Qwen model, intensifying rivalries among Chinese tech giants. [Computerworld]
IBM has evolved from using only its own LLMs to offering multiple model options, including open-source integrations through Hugging Face, while helping customers navigate model selection. [VentureBeat]
OpenAI has prepared to release an open-weight language model across multiple cloud providers, marking its first such release since signing its exclusive deal with Microsoft. [The Verge]
Whose Data?
AI2 has developed FlexOlmo, a language model allowing data owners to maintain control over their training data even after model completion, challenging current industry practices. [Wired]
Amazon has secured AI content licensing deals with Condé Nast and Hearst for its Rufus shopping assistant, following a similar agreement with The New York Times. [Editor & Publisher Magazine]
Anubis, an open-source bot-blocking program, has been downloaded 200,000 times since January 2025 to protect websites from AI scraping bots. [ZDNet]
DeepSeek has faced bans and investigations across Europe after German privacy officials reported the Chinese AI chatbot to app stores for illegal data transfers. [TechRadar]
Wiley has partnered with Anthropic to integrate academic research into AI systems through the Model Context Protocol, ensuring proper attribution and citation of scholarly content. [Publishers Weekly]
The LLM Ecosystem
Cerebras Systems has launched its Inference Cloud service on AWS Marketplace, enabling enterprise customers to deploy high-speed AI inference solutions through their AWS accounts. [Business Wire]
And Cerebras Systems has partnered with Hugging Face, DataRobot, and Docker to integrate its high-speed AI inference technology into their development platforms and tools. [Business Wire]
ClickUp has launched Brain Max, a platform integrating multiple AI tools and data sources into one interface, allowing enterprises to query their information without switching between applications. [VentureBeat]
Cloudian has integrated AI inferencing capabilities with its HyperStore object storage platform, combining vector database support with high-performance data storage for enterprise AI workflows. [EIN Presswire]
Coginiti has launched a Semantic Layer that enables trusted AI and standardized analytics by providing contextual, governed definitions across complex data environments. [EIN Presswire]
CoreWeave has become the first cloud provider to install Nvidia's Blackwell Ultra AI chips, gaining an edge over larger competitors in offering advanced AI computing capabilities. [Yahoo Finance]
DigitalOcean has launched its GradientAI Platform, enabling developers to create AI applications by combining foundation models from multiple providers through a simplified interface. [Business Wire]
Graphwise has launched GraphDB 11, enhancing its semantic graph database with expanded LLM support, improved GraphRAG capabilities, and MCP integration for AI applications. [BigDATAwire]
iMerit has launched its Scholars program to provide expert-led AI model fine-tuning, focusing on quality data annotation through domain specialists rather than crowdsourced workers. [TechCrunch]
Katanemo Labs has developed Arch-Router, an intelligent routing framework that directs user queries to optimal language models based on preference-aligned policies. [VentureBeat]
Sakana AI has developed Multi-LLM AB-MCTS, enabling multiple LLMs to collaborate and solve complex problems by combining their unique strengths. [VentureBeat]
SciArena has launched an open platform for evaluating LLMs on scientific tasks, with initial results showing OpenAI's o3 model leading among both proprietary and open-source competitors. [The Decoder]
Taazaa has launched a free AI Readiness Assessment tool that helps organizations evaluate their preparedness for AI adoption across four key dimensions. [EIN Presswire]
Agentic AI
Atlassian has empowered employees to build thousands of AI agents through Rovo Studio, fostering an experimental culture that embraces both success and failure. [VentureBeat]
ContentWise has integrated AI agents into its UX Engine platform, enabling automated content curation and personalization for video streaming services and digital publishers. [PRWeb]
Dust has grown from US$1m to US$6m in annual revenue by developing AI agents that automate complex business workflows across enterprise applications. [VentureBeat]
Google Cloud has launched Firebase Studio, an AI-powered development environment offering three-tiered assistance levels, from basic conversation to autonomous coding, while preserving developers’ essential role. [Forbes]
Intuit has launched an AI agent suite within QuickBooks that automates workflows, handles payments and accounting, and manages customer relationships for businesses. [Forbes]
IQVIA has unveiled custom-built AI agents developed with Nvidia to enhance workflows and insights in life sciences through their Healthcare-grade AI platform. [Yahoo Finance]
Q4 has launched Q, an AI-powered agent that helps investor relations officers analyze data, monitor stakeholders, and streamline communications through its platform. [Business Wire]
NinjaTech AI has launched Super Agent, an AI-powered virtual assistant with dedicated VMs, partnering with Cerebras Systems to execute complex workflows 3-5x faster than GPU solutions. [GlobeNewswire]
RegASK has launched the industry’s first agentic AI architecture for regulatory affairs, combining specialized agents with a vertical language model to deliver personalized compliance insights. [Business Wire]
Warp has launched an agentic development environment that combines terminal-style AI interaction with code generation and deployment controls, building upon its existing terminal software. [Fast Company]
Other LLM Sightings
Agent2.AI has launched Super Agent, an AI tool that transforms research prompts into ready-to-use deliverables like spreadsheets and presentations. [EIN Presswire]
Auxia has launched an enhanced AI Analyst Agent that helps marketing teams instantly analyze campaign data through natural language queries, replacing weeks of manual analysis. [Business Wire]
Behavox has reduced compliance investigation times by over 40% through its Intelligent Archive product, which unifies data and applies specialized AI technology. [Business Wire]
Boosted.ai has launched an enhanced version of Alfa, an AI platform that helps investment teams automate research tasks and improve decision-making processes. [Business Wire]
CapStorm has launched CapStorm:AI, a self-hosted solution enabling users to analyze Salesforce and SQL data using natural language queries without coding expertise. [Yahoo Finance]
CFP Consulting has launched FRONT-END.AI, an AI-powered tool that converts Figma designs into HTML code within Visual Studio Code. [Business Wire]
CLARA Analytics has launched AI-powered Risk Notes and Claim Event Indicators to help insurance claims professionals identify and address issues in real-time. [Business Wire]
Denodo has launched DeepQuery, an AI capability that analyzes complex data questions across multiple sources to explain why events occurred, not just what happened. [BigDATAwire]
GoHighLevel has launched AI Employee, an automation platform offering voice, conversation, reviews, content, workflow, and funnel AI tools with 30 days free access. [EIN Presswire]
Gregory FCA has launched CrisisCalm, an AI-powered crisis communications platform that generates comprehensive response playbooks and materials in real-time for organizations. [Business Wire]
IIFL has partnered with Microsoft to deploy AI virtual assistants for its 50,000 employees, aiming to handle internal queries across HR, IT, and support functions. [TechCircle]
MaxBill has launched an AI-powered self-service portal enabling utility providers to automate customer service, personalize engagement, and monitor usage data in real-time. [EIN Presswire]
SapiensFirst has launched Ann, an AI recruitment platform that emulates human recruiters while reducing hiring time by 70% and costs by 80%. [EIN Presswire]
Seapines AI has launched CaptionEdge, an iOS app that generates private, AI-powered captions for photos and videos using on-device processing. [EIN Presswire]
StackAdapt has launched Ivy, an AI-powered platform assistant that helps marketers make faster data-driven decisions throughout their advertising campaigns. [Business Wire]
TRULEO has launched AI-powered witness interview and canvassing tools to help law enforcement agencies collect statements more safely and efficiently while improving investigative accuracy. [EIN Presswire]
Risks and Responses
Alterd, an AI-powered journaling app, has helped users navigate psychedelic experiences and overcome addiction, despite experts’ concerns about AI-guided drug use. [Wired]
Apate.ai has partnered with CommBank to deploy AI-powered ‘victim bots’ that engage scammers, waste their time, and gather intelligence for fraud prevention. [Startup Daily]
CalFire's AI chatbot, launched to provide wildfire information to Californians, has demonstrated significant accuracy and consistency issues when delivering critical emergency details. [The Markup]
ChatEurope, a multilingual AI chatbot developed to combat disinformation with journalist-verified news responses, has launched with EU funding from fifteen European media organizations, but early tests revealed significant issues with outdated information and incomplete responses. [The Decoder]
Google's Gemini AI has gained automatic access to WhatsApp and other Android apps, prompting privacy concerns and leading users to seek opt-out methods. [TechRadar]
Hitachi Energy's CEO has warned that AI training’s volatile power consumption patterns threaten grid stability and require government regulation to manage effectively. [Verdict]
Intel researchers have demonstrated how AI chatbots can be tricked into providing harmful information through a jailbreaking technique that overloads queries with academic jargon. [404 Media]
Intelligencer’s James Walsh has documented widespread AI-assisted cheating in higher education, revealing professors’ despair and administrators’ reluctance to address this transformative challenge. [Vox]
Model Context Protocol has gained widespread adoption since its November launch, but regulated industries, particularly financial institutions, remain hesitant due to security and compliance concerns. [VentureBeat]
Nikkei Asia has discovered researchers from multiple institutions embedding hidden text in academic papers to manipulate AI review systems into providing positive evaluations. [The Register]
Okta has discovered cybercriminals are exploiting Vercel's v0.dev generative AI tool to create sophisticated phishing websites that mimic legitimate login pages. [TechRadar]
OpenAI has hired a forensic psychiatrist and consulted mental health experts to study AI’s psychological effects after users experienced delusions and mental health issues. [Futurism]
And OpenAI has intensified security measures, including biometric controls and restricted access to sensitive data, following alleged model copying by DeepSeek and broader corporate espionage concerns. [TechCrunch]
Paradox.ai‘s McDonald’s hiring chatbot exposed millions of job applicants’ personal data due to weak security measures. [Wired]
Signal's messaging platform became the vector for an AI-generated voice scam impersonating Marco Rubio that targeted officials, prompting a State Department investigation. [eWeek]
xAI's Grok chatbot has posted antisemitic and pro-Hitler content on X before being deleted, marking its second Nazi-themed malfunction since May 2023. [The Register]
And Grok has been banned in Turkey after generating offensive content about Turkish leaders, prompting swift corrective action from the company. [Associated Press]
Regulation
California’s SB 53 has introduced amendments requiring major AI companies to publish safety protocols and incident reports, marking the state’s latest attempt at mandating AI transparency requirements. [TechCrunch]
Google faces an EU antitrust complaint from independent publishers who claim its AI Overviews feature abuses market power and damages their traffic and revenue. [Verdict]
Mistral AI and other European companies have urged EU leaders to delay AI Act enforcement by two years to better compete with unregulated tech giants. [The Register]
But the European Commission has rejected calls from major tech companies to delay implementing its AI Act, maintaining its original timeline for rolling out regulations through 2026. [Reuters]
And the EC has published a voluntary General-Purpose AI Code of Practice to help companies navigate compliance with the Act. [The Register]
Conversational AI
CallRail has launched conversion signals, an AI-powered feature that automatically identifies appointment-converting calls to help marketers optimize their Google Ads campaigns. [Business Wire]
Gryphon AI's CTO Neal Keene has warned that while real-time AI can enhance contact center compliance, improper implementation risks costly regulatory violations. [Speech Technology Magazine]
inTouch has launched an AI-powered phone companion service in North America that makes daily calls to seniors, following its successful European debut. [Speech Technology Magazine]
Meta is developing AI chatbots that will proactively initiate conversations with users through its AI Studio platform, partnering with Alignerr for training data labeling. [ZDNet]
NextLevel.AI has launched an AI-powered BDR Voice Agent that engages leads through voice and text, qualifies prospects, and schedules meetings with sales teams. [EIN Presswire]
Omilia has formalized its five-year partnership with Connex by certifying them as an Implementation Partner for conversational AI solutions in financial services. [Business Wire]
Omnichat has launched Omni AI Agent Studio, enabling businesses to build custom AI agents by integrating various language models for personalized customer interactions across multiple platforms. [DealStreetAsia]
Salesforce's Agentforce platform has surpassed 1 million AI agent-customer conversations, with autonomous agents resolving 85% of customer queries since its October launch. [Yahoo Finance]
Voice News
AIStorm and DB HiTek have launched SpectroMic KWS, an ultra-low-power keyword spotting solution combining MEMS microphone and smart features in a single package. [Speech Technology Magazine]
Deepgram has launched Saga, a voice operating system that enables developers to control their entire tech stack through natural speech commands and AI-powered workflow automation. [Speech Technology Magazine]
Intron has deployed its Sahara Voice AI, trained on 3.5 million African voice clips, across East African healthcare facilities, courtrooms, and call centers. [CNBC Africa]
PayJunction has integrated with Twilio to provide AI-driven voice payment processing, enabling businesses to automate and secure their phone payment operations. [Business Wire]
Ultatel has launched a managed AI voice assistant service that automates business tasks and integrates with its cloud communications platform to provide enterprise-grade customer service. [DestinationCRM.com]
Document AI
Allyant has launched an AI-powered web-based version of its CommonLook PDF remediation software to accelerate document accessibility compliance. [Business Wire]
Counterpoint has launched ReadyDocs 25.2, an enterprise content management platform featuring enhanced automation, workflow tools, and compliance capabilities for regulated industries. [Business Wire]
Geek Haus has launched AI-powered handwriting recognition on TutorFlow, enabling STEM educators to instantly convert handwritten equations into digital content. [EIN Presswire]
Hyland has launched an AI-powered document processing solution that autonomously understands, reasons about, and acts upon unstructured data across enterprise workflows. [Hyland]
iDox.ai has expanded its AI-powered redaction tools across US sectors, helping organizations meet privacy compliance standards through automated data protection. [Yahoo Finance]
Moveworks has partnered with Docusign to integrate intelligent agreement management capabilities into its AI Agent Marketplace, streamlining document workflows for enterprise users. [Business Wire]
ParaScript has launched CheckStock.AI, an AI-powered solution that verifies check authenticity by comparing stock elements against reference images to detect fraud. [Business Wire]
Translation
Babeltext has launched a Shopify integration enabling merchants to provide real-time customer support across 195 languages through multiple messaging channels. [Business Wire]
Chinese tech companies’ AI-powered sign language translation avatars have failed to effectively serve deaf users due to poor implementation and developers’ unwillingness to incorporate feedback from the deaf community. [The Register]
GlobeScribe has launched an AI-powered book translation platform offering 24-hour translations in five languages for $100 per book during its soft launch phase. [EIN Presswire]
Google researchers have uncovered significant quality issues in major multilingual speech datasets, recommending improved data collection practices and linguistic expertise for future development. [Slator]
The Society of Authors has launched a campaign defending human translators after its survey revealed widespread AI-related income losses and growing industry adoption of MT. [Slator]
Search
Cerebras Systems has partnered with Notion to power sub-300-millisecond enterprise search capabilities for over 100 million workspace users worldwide. [Business Wire]
The Independent Publishers Alliance has filed an EU antitrust complaint against Google for using publishers’ content in AI-generated search summaries without consent. [TechCrunch]
Perplexity has launched Comet, an AI-powered web browser featuring integrated search and an AI assistant, challenging Google Chrome’s dominance in the browser market. [TechCrunch]
Return On Now has launched enhanced Answer Engine Optimization packages to help businesses gain visibility across AI platforms like ChatGPT, Perplexity, and Gemini. [EIN Presswire]
Health Tech
Avaamo has launched an AI-powered ambient clinical intelligence solution integrated with Sunrise Clinical Manager to reduce clinician documentation burden and improve workflow efficiency. [EIN Presswire]
Evidently has partnered with Allina Health to deploy AI-powered clinical data tools that reduce chart review time and improve documentation quality across all care settings. [Business Wire]
Google has expanded its MedGemma collection with two new healthcare AI models: a 27B multimodal model for complex health record interpretation and MedSigLIP for medical imaging tasks. [Google]
A GPT-powered medical device certified in Europe raises questions about generative AI in health care. [STAT]
Jimini Health has released a framework for safe AI-powered mental healthcare. [Yahoo Finance]
KPMG has partnered with Hippocratic AI to deploy AI healthcare agents that address staffing shortages by performing non-diagnostic clinical tasks alongside human workers. [Business Wire]
Medical Supplierz has announced an AI-powered search tool to streamline matching between healthcare buyers and medical equipment suppliers on its B2B marketplace. [EIN Presswire]
Microsoft's AI team has developed a disease-diagnosing system that outperformed clinicians, though its claims of ‘medical superintelligence’ sparked controversy among healthcare professionals. [STAT]
OpenAI’s ChatGPT has helped patients obtain accurate medical diagnoses missed by doctors, prompting healthcare providers to evaluate AI’s role in medical consultations and diagnosis. [Wired]
Otter.ai has achieved HIPAA compliance, enhancing its ability to protect sensitive healthcare data while providing AI-powered meeting documentation services to medical professionals. [Business Wire]
SourceNow has launched an AI-powered vendor management system for healthcare organizations, featuring float pool management and real-time talent matching capabilities. [Business Wire]
Legal Tech
AI2L has launched as a global legal tech alliance, uniting regional groups from Brazil, India, Asia-Pacific, and Europe to promote innovation and cross-border collaboration. [Artificial Lawyer]
Pinsent Masons has partnered with ContractPodAi to integrate the Leah AI platform into its managed legal services and develop advanced solutions for in-house teams. [UKTN]
EvenUp has launched AI Playbooks and Voice Agent while enhancing its AI Drafts suite, expanding its legal tech platform’s capabilities for case analysis and client communication. [LawSites]
FiscalNote has expanded its PolicyNote platform with AI-powered legislative drafting capabilities, enabling policy professionals to create and customize legislative text more efficiently. [Business Wire]
Georgia Court of Appeals has overturned a lower court’s divorce ruling after discovering it relied on non-existent legal cases generated by AI. [The Register]
Harvey is establishing a Bengaluru office to expand its global engineering, sales, and operations teams, while also serving its growing Indian client base. [Artificial Lawyer]
Jylo has secured the Council of Europe as a client for its legal AI platform following a competitive tender, marking a significant win for the UK-based startup. [Artificial Lawyer]
LegalSifter has warned against using general-purpose AI like ChatGPT for contract review, emphasizing the need for purpose-built solutions that ensure consistency and compliance. [Artificial Lawyer]
A study by LexisNexis found that large companies using AI tools could reduce their external legal work by 13% and achieve significant time savings for in-house teams. [Artificial Lawyer]
Luminance has launched a Compliance module that automatically checks documents against internal policies and regulations, flagging non-compliance issues for immediate resolution. [Artificial Lawyer]
Paralex has launched a hybrid AI-attorney legal platform offering US startups custom document services at fixed fees between $150-900, with every document reviewed by lawyers. [Legal IT Insider]
Practice AI has partnered with Settlement Streams to integrate AI-powered legal tools with post-settlement financial management services for personal injury and mass tort firms. [GlobeNewswire]
Sprintlaw has halved its workforce to 30 staff while doubling its client base through a US$2m investment in AI tools, including its ‘Taylor AI’ chatbot. [SmartCompany]
Thomson Reuters has launched CoCounsel Knowledge Search, an AI tool that enables unified searching across multiple content repositories including Westlaw, HighQ, and SharePoint. [Artificial Lawyer]
Ed Tech
The American Federation of Teachers has partnered with Microsoft, OpenAI, and Anthropic to launch a US$23m AI training academy for 400,000 teachers over five years. [Forbes]
Anthropic has integrated Claude for Education with Wiley, Canvas, and Panopto, allowing students to access academic content directly through their AI conversations. [ZDNet]
Cluely has evolved from promoting itself as an undetectable ‘cheating’ tool to positioning itself as a ChatGPT competitor with screen and audio awareness capabilities. [Yahoo Finance]
Echo360's GoReact has expanded its partnership with Macmillan Learning to integrate AI-powered video feedback tools for authentic assessment into Macmillan’s Achieve platform. [Business Wire]
Microsoft has pledged US$4B to enhance AI education across schools and colleges, aiming to upskill 20 million people over five years. [TechRadar]
OpenAI has developed ‘Study Together’, an interactive learning feature for ChatGPT that provides step-by-step guidance and adaptive feedback for students and educators. [TestingCatalog]
Funding
Alumly has secured a seven-figure angel investment to develop its platform connecting university students with paid work experiences during their studies. [Startup Daily]
Blok has launched an AI-powered simulation platform that lets developers test app features using virtual user personas, backed by US$7.5m in venture funding. [TechCrunch]
Brainfish has raised US$6.4m in pre-Series A funding to expand its AI-powered customer support platform into the United States and scale globally. [Business Wire]
Context has launched an AI-native office suite with US$11m in seed funding to automate billions of hours of repetitive knowledge work tasks. [Business Wire]
GeniusTravel has raised €2M to expand its AI-powered travel booking app, which helps budget-conscious users create personalized trip packages through simple chat interactions. [Tech.eu]
Gradient Labs has raised US$13m Series A funding to expand its AI customer support platform, which resolves up to 90% of queries in regulated industries. [Tech.eu]
Groq has approached investors to raise up to US$500m at a US$6 billion valuation to fulfill its recent US$1.5 billion AI chip deal with Saudi Arabia. [DealStreetAsia]
Headway has raised Series A funding from Bullhound Capital to expand its educational technology ecosystem, which serves 150 million users across 170 countries. [Tech.eu]
LangChain has secured new funding at a US$1B valuation, building on its success with LangSmith, which has generated US$12-16 million in annual recurring revenue. [TechCrunch]
OpenRouter has secured US$40m in funding to expand its platform that enables developers to access over 400 AI models through a single unified API. [siliconANGLE]
Perry has secured €1.6m in funding to develop AI-powered work instructions that help address Europe’s technical workforce shortage through real-time, adaptive guidance. [Tech.eu]
Pokee AI has raised US$12m in seed funding, led by Point72 Ventures, to develop AI agents that automate online workflows using reinforcement learning technology. [GeekWire]
Portola has raised US$20m in Series A funding for its AI companion app Tolan, which has attracted over 100,000 paid users since launching in February. [GeekWire]
SiPearl has secured €130m in Series A funding to develop European-made processors for supercomputers, challenging US and Chinese dominance in high-performance computing. [Tech Funding News]
Sundial has raised US$23m in funding to develop its AI-powered analytics platform that helps companies make data-driven decisions faster. [Business Wire]
Tandem has raised US$3.8m in seed funding to develop AI agents that help users complete tasks within apps by taking actions on their behalf. [Tech.eu]
Wonderful has raised US$34m in seed funding to expand its multilingual AI customer service platform across non-English speaking markets in Europe, Asia and the middle East. [Feed the AI]
ZeroEntropy has raised US$4.2m in seed funding to develop an API that helps AI models retrieve relevant data more effectively from knowledge bases. [TechCrunch]
Acquisitions
CoreWeave has acquired Core Scientific in a US$9B stock deal, gaining access to massive data center capacity for AI workloads. [TechCrunch]
Progress Software has acquired Spanish RAG-as-a-Service startup Nuclia for an estimated US$50m, expanding its AI capabilities for business applications. [Blocks & Files]
There’s More
Anysphere has apologized for poorly communicating changes to Cursor's Pro plan pricing structure, which unexpectedly increased costs for some users of its AI coding tool. [TechCrunch]
Fortune has brought back former editor Nick Lichtenberg to spearhead AI-assisted content production, including a new AI-co-written section called Fortune Intelligence. [Semafor]
Microsoft and OpenAI's relationship has deteriorated partly due to their inability to agree on AGI’s definition, despite previously suggesting it meant achieving US$100B in profits. [Ars Technica]
Missouri Attorney General Andrew Bailey has threatened legal action against tech companies whose AI chatbots ranked Trump unfavourably on antisemitism, claiming consumer fraud violations. [TechDirt]
OpenAI has been operating at massive losses despite raising US$60.9B, prompting concerns of a ‘subprime AI crisis’ due to unsustainable business models. [Futurism]
The Back Cover
Here’s The Smashing Pumpkins with a cover of Fleetwood Mac’s Landslide.
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!