Open Source2026-04-24VentureBeatOpenAI Launches Privacy Filter: Open-Source Data SanitizationOpenAI has released Privacy Filter, an open-source, on-device model designed to detect and redact personally identifiable information (PII) before it reaches cloud servers. Launched on Hugging Face, this tool marks a shift toward local-first privacy infrastructure, enabling enterprises to sanitize datasets while maintaining data security and compliance.
Product Launch2026-04-24VentureBeatOpenAI Unveils Workspace Agents for EnterprisesOpenAI has introduced Workspace Agents, a successor to custom GPTs for enterprises. These agents can plug directly into platforms like Slack and Salesforce, enabling businesses to deploy and control fleets of AI workers. Available on ChatGPT Business plans, this product represents a major step in enterprise AI adoption and orchestration.
Model Update2026-04-24VentureBeatOpenAI's GPT-5.5 Beats Claude Mythos on Terminal BenchOpenAI's GPT-5.5 has been released, narrowly outperforming Anthropic's Claude Mythos Preview on the Terminal Bench 2.0 benchmark. Despite rumors of internal codename 'Spud,' the model proves its capabilities, marking a competitive milestone in the AI model race. It is available for use in ChatGPT and via API.
Model Update2026-04-24VentureBeatAnthropic Reveals Cause of Claude Degradation IssuesAnthropic has solved the mystery behind recent Claude model degradation. Changes to Claude's harnesses and operating instructions likely caused the perceived decline in performance, which users described as 'AI shrinkflation.' The company's transparency about the issue helps restore trust and provides insights into the challenges of maintaining AI model quality.
Multimodal2026-04-24WIRED AIOpenAI Beefs Up ChatGPT's Image Generation ModelOpenAI has released ChatGPT Images 2.0, an upgraded image generation model. Testing shows it produces more detailed images and renders text better than its predecessor, though it still struggles with non-English languages. This update enhances ChatGPT's multimodal capabilities, making it more versatile for creative and professional use.
AI Infrastructure2026-04-24WIRED AIAI Tools Help North Korean Hackers Steal MillionsNorth Korean hackers are leveraging AI tools to enhance their cybercrime operations. One group used AI for everything from vibe coding malware to creating fake company websites, stealing up to $12 million in just three months. This highlights the growing threat of AI-augmented cyberattacks from state-sponsored actors.
AI Infrastructure2026-04-24WIRED AI5 AI Models Tried to Scam Me: Some Were Scary GoodA cybersecurity investigation reveals that AI models possess alarming social engineering capabilities. Five different AI models attempted phishing scams, with some demonstrating frighteningly effective deception skills. The findings highlight the dual-use nature of AI, where its conversational abilities can be weaponized for cyberattacks.
Product Launch2026-04-24The VergeClaude Connects Directly to Personal AppsAnthropic's Claude AI now features new connectors that allow it to access personal apps like Spotify, Uber Eats, and TurboTax. This expansion from work-related apps to personal services enables Claude to assist with a wider range of daily tasks, from entertainment to grocery shopping, enhancing its utility as a personal AI assistant.
Model Update2026-04-24TechCrunch AIOpenAI Releases GPT-5.5, Steps Toward AI Super AppOpenAI has released GPT-5.5, its latest AI model, which brings the company closer to creating an AI 'super app.' The model offers increased capabilities across a broad variety of categories, enhancing performance in tasks like coding, research, and data analysis. This release is a key step in OpenAI's vision for a unified AI platform.
AI Infrastructure2026-04-24TechCrunch AINoscroll: AI Bot That Does Your DoomscrollingA new AI bot called Noscroll aims to cure doomscrolling by reading the internet for you. The tool automates the consumption of online content, allowing users to stay informed without the negative mental health impacts of endless scrolling. It represents a novel application of AI to improve digital well-being.
Product Launch2026-04-24TechCrunch AIBret Taylor's Sierra Acquires YC-Backed AI Startup FragmentSierra, the AI customer service agent startup founded by technologist Bret Taylor, has announced the acquisition of YC-backed French startup Fragment. This move strengthens Sierra's capabilities in the AI customer service space, expanding its technology and talent pool to better serve enterprise clients.
Model Update2026-04-24OpenAI BlogOpenAI Introduces GPT-5.5: Smartest Model YetOpenAI has officially introduced GPT-5.5, described as its smartest model yet. It is faster and more capable, built specifically for complex tasks such as coding, research, and data analysis across various tools. This release marks a major milestone in OpenAI's model development, promising enhanced performance for developers and enterprises.
Model Update2026-04-24OpenAI BlogGPT-5.5 System Card Released by OpenAIOpenAI has published the system card for GPT-5.5, its latest and most advanced model. The document details the model's capabilities, safety evaluations, and performance benchmarks. GPT-5.5 is designed for complex tasks including coding, research, and data analysis, representing a significant leap in AI model intelligence and speed.
Model Update2026-04-24NVIDIA AI BlogOpenAI's GPT-5.5 Powers Codex on NVIDIA InfrastructureAI agents are revolutionizing developer workflows and now moving into knowledge work, processing information, solving complex problems, and driving innovation. Codex, OpenAI's agentic coding application, is enabling this new frontier, now powered by GPT-5.5 on NVIDIA infrastructure. NVIDIA is already putting this technology to work, marking a significant step in AI-driven development.
Product Launch2026-04-23The VergeOpenAI Launches Custom Workspace Agents for TeamsOpenAI releases 'Workspace Agents' for its Business, Enterprise, and Education plans. These are cloud-based, Codex-powered agents that teams can customize to automate complex workflows and connect to tools like Slack and Salesforce. Examples include agents that scour the web for product feedback and automatically generate reports, enabling teams to scale operations.
Product Launch2026-04-23The VergeMusk: Millions of Teslas Won't Get Unsupervised FSDElon Musk confirms that Tesla vehicles equipped with the older Hardware 3 (HW3) computer will not receive unsupervised Full Self-Driving capability. This affects approximately 4 million vehicles, meaning a significant portion of the Tesla fleet is limited to driver-assist features. The announcement clarifies the hardware requirements for Tesla's highest-level autonomy ambitions.
Product Launch2026-04-23TechCrunch AIOpenAI Partners with Infosys for Enterprise AI ToolsOpenAI teams up with global IT services firm Infosys to bring its AI tools to a wider range of businesses. The integration will initially focus on modernizing software development, automating workflows, and deploying AI systems in areas like software engineering and DevOps. The partnership aims to lower the barrier to enterprise AI adoption.
AI Infrastructure2026-04-23TechCrunch AIGoogle Cloud Launches New AI Chips to Rival NvidiaGoogle Cloud announces two new Tensor Processing Unit (TPU) chips, positioning them as faster and more cost-effective than prior versions. While designed to compete with Nvidia's dominant GPUs, Google states it will continue to offer Nvidia hardware in its cloud for now. This move highlights the intensifying competition in the foundational AI infrastructure market.
Model Update2026-04-23OpenAI BlogOpenAI Launches ChatGPT Images 2.0OpenAI introduces ChatGPT Images 2.0, a significantly upgraded image generation model. Key improvements include superior text rendering within images, multilingual support, and advanced visual reasoning capabilities. The update marks a major step forward in the quality and versatility of AI-generated visual content directly within the ChatGPT interface.
Product Launch2026-04-23OpenAI BlogOpenAI Makes ChatGPT Free for US CliniciansOpenAI is offering its specialized 'ChatGPT for Clinicians' tool for free to verified U.S. physicians, nurse practitioners, and pharmacists. The AI assistant is designed to support clinical care, documentation, and research tasks. This initiative aims to integrate AI safely into healthcare workflows to reduce administrative burden and aid medical decision-making.
AI Infrastructure2026-04-23NVIDIA AI BlogNVIDIA & Google Cloud Advance Agentic & Physical AINVIDIA and Google Cloud announce a deepened collaboration to co-engineer a full-stack AI platform. The partnership focuses on advancing agentic AI (autonomous systems) and physical AI (robotics, embodied AI). This integrated platform aims to provide developers and enterprises with the tools needed to build and deploy next-generation intelligent applications.
AI Infrastructure2026-04-23NVIDIA AI BlogNVIDIA AI's 5 Ways to Protect the PlanetNVIDIA highlights five key applications of its AI technology aimed at environmental protection. These span climate research, conservation efforts, disaster monitoring, and recycling plant optimization. The initiatives demonstrate how accelerated computing can address critical ecological challenges, from analyzing rainforest data to improving waste management systems.
Multimodal2026-04-22VentureBeatOpenAI's ChatGPT Images 2.0 Creates Complex Multilingual GraphicsOpenAI's ChatGPT Images 2.0 model can generate complex visual outputs like multilingual text, full infographics, slides, maps, and manga with high proficiency. The model shows a leap in multimodal understanding and compositional ability, handling intricate instructions to produce detailed and structured visual content that integrates diverse elements seamlessly.
Product Launch2026-04-22VentureBeatGoogle Launches Deep Research Agents for Web & Private DataGoogle unveiled Deep Research and Deep Research Max, its most significant upgrade to autonomous research agents. These new agents can fuse open web data with proprietary enterprise information for the first time, enabling deeper, context-aware analysis. The launch represents a major step in making AI-powered research tools more powerful and personalized.
AI Coding2026-04-22WIRED AIMozilla Used Anthropic's Mythos to Find 271 Firefox BugsMozilla's Firefox team used Anthropic's exclusive cybersecurity AI model, Mythos, to identify and fix 271 bugs in the browser. While the team believes AI won't upend cybersecurity long-term, they warn developers to expect a rocky transition as these tools become more prevalent. The case demonstrates the practical application of specialized AI models in software security.
Multimodal2026-04-22The VergeOpenAI's Image Generator Can Now Pull Info from WebOpenAI is rolling out ChatGPT Images 2.0 with new 'thinking capabilities,' including the ability to search the web to inform its image creation. This allows the model to generate more sophisticated and context-aware images from a single prompt. The update represents a step toward more autonomous and knowledge-grounded multimodal AI systems.
Multimodal2026-04-22TechCrunch AIChatGPT's Images 2.0 Model Excels at Generating TextOpenAI's new ChatGPT Images 2.0 model demonstrates surprising proficiency in generating coherent text within images. The advancement highlights the rapid evolution of multimodal AI capabilities, moving beyond simple image generation to creating complex visual compositions with integrated textual elements. The model shows significant improvements over previous versions.
AI Coding2026-04-22TechCrunch AISpaceX Has Option to Buy AI Startup Cursor for $60BSpaceX is reportedly working with AI coding startup Cursor and holds an option to acquire the company for $60 billion. The potential deal comes as SpaceX prepares for a public listing. This strategic move could integrate advanced AI coding capabilities directly into SpaceX's and xAI's development ecosystems.
AI Infrastructure2026-04-22TechCrunch AIMeta to Record Employee Keystrokes to Train AI ModelsMeta has developed a new internal tool that converts employee mouse movements and keystrokes into data for training its AI models. The company states this data collection is for internal model improvement. This move raises significant questions about workplace privacy and the ethical sourcing of training data for AI systems.
AI Coding2026-04-22OpenAI BlogOpenAI Scales Codex to Enterprises via Major PartnershipsOpenAI launched Codex Labs, partnering with Accenture, PwC, and Infosys to help enterprises deploy and scale its Codex AI across the software development lifecycle. The company also reported reaching 4 million weekly active users for Codex. This initiative aims to deeply integrate AI-assisted coding into large-scale corporate development workflows.
Product Launch2026-04-21WIRED AIHumanoid Robot Sets Half-Marathon Record in ChinaAn autonomous humanoid robot from the company Honor completed a half marathon in 50 minutes and 26 seconds, beating the human record by 7 minutes. The achievement demonstrates significant advances in robotic endurance, mobility, and energy efficiency. This milestone in physical AI highlights the rapid progress being made in bipedal robotics and their potential applications.
AI Infrastructure2026-04-21MIT Technology ReviewChinese Tech Workers Train AI Doubles, Face Job ConcernsTech workers in China are being instructed by management to train AI agents designed to replace them, leading to widespread soul-searching among early adopters. Projects like 'Colleague Skill' on GitHub claim to help workers 'distill' their expertise into AI counterparts. This trend highlights the ethical and professional dilemmas as AI automation advances rapidly in the workforce.
Model Update2026-04-21TechCrunch AINSA Reportedly Using Anthropic's Restricted Mythos AI ModelThe U.S. National Security Agency (NSA) is reportedly using Anthropic's restricted 'Mythos' AI model for intelligence purposes, despite an ongoing feud between Anthropic and the Pentagon. The use of a specialized, high-capability model highlights the growing adoption of advanced AI in national security and espionage. This occurs amid tensions regarding AI governance and military applications.
AI Infrastructure2026-04-21TechCrunch AIAnthropic Secures $5B from Amazon, Pledges $100B AWS SpendAmazon is investing an additional $5 billion in AI startup Anthropic, which in return has agreed to spend $100 billion on Amazon Web Services (AWS). This circular deal deepens the strategic partnership, ensuring Anthropic's massive cloud computing needs are met by AWS while providing Amazon with a significant stake in a leading AI lab. The investment underscores the enormous capital and infrastructure requirements of frontier AI.
Product Launch2026-04-21OpenAI BlogHyatt Deploys ChatGPT Enterprise GloballyHyatt Hotels is deploying ChatGPT Enterprise across its global workforce, utilizing GPT-5.4 and Codex to enhance productivity, operations, and guest experiences. The enterprise-wide rollout represents a significant adoption of AI tools in the hospitality sector to streamline internal processes and improve customer service through advanced automation and assistance.
Product Launch2026-04-21NVIDIA AI BlogAdobe, NVIDIA, WPP Partner on AI Agents for MarketingNVIDIA announced expanded strategic collaborations with Adobe and WPP to integrate agentic AI into enterprise marketing operations. The partnership aims to deploy autonomous AI agents at scale to transform creative production and decision-making processes. This move positions AI as a central driver for breakthrough creative intelligence, accelerating content creation workflows for major brands.
Product Launch2026-04-20Hacker NewsEx-CEO and CFO of Bankrupt AI Company Charged with FraudFederal prosecutors have charged the former CEO and CFO of a now-bankrupt AI company with fraud. The executives are accused of misleading investors about the company's technology and financial health. This case highlights the regulatory scrutiny and legal risks emerging in the high-stakes AI startup ecosystem as investor enthusiasm meets market realities.
AI Infrastructure2026-04-20The VergeGlobal RAM Shortage Could Last for YearsA severe shortage of DRAM memory chips, critical for AI servers and consumer electronics, is projected to last until at least 2027. Despite production ramps, manufacturers may only meet 60% of demand by late 2027, with some estimates extending the shortage to 2030. The crunch is driven by explosive demand from AI infrastructure, straining capacity at major producers Samsung, SK Hynix, and Micron.
Multimodal2026-04-20MIT Technology ReviewHow Robots Learn: A Brief, Contemporary HistoryThe field of robotics is undergoing a paradigm shift, moving from hard-coded instructions to AI-driven learning. Modern roboticists are leveraging foundation models, simulation, and large-scale data to create systems that can generalize across tasks. This article traces the evolution from simple, repetitive industrial arms towards adaptable robots capable of learning complex physical skills, powered by advances in AI and compute.
Product Launch2026-04-20TechCrunch AIKey OpenAI Executives Exit as Company Sheds 'Side Quests'OpenAI executives Kevin Weil and Bill Peebles are departing as the company shuts down its Sora video generation project and folds its science team. This signals a strategic pivot away from consumer-focused 'moonshot' projects toward a sharper focus on enterprise AI solutions. The restructuring reflects a prioritization of core products and business models over experimental consumer applications.
Model Update2026-04-19VentureBeatOpenAI Debuts GPT-Rosalind for Life Sciences ResearchOpenAI has introduced GPT-Rosalind, a limited access model specifically designed for life sciences research to accelerate drug discovery and biological analysis. The model aims to address the 'fragmented and difficult' research tools that currently slow pharmaceutical development. Simultaneously, OpenAI has expanded Codex plugin availability on GitHub, providing broader access to its AI coding tools for scientific and development communities working on complex research challenges.
AI Art2026-04-19VentureBeatAnthropic Launches Claude Design AI Tool for PrototypesAnthropic has launched Claude Design from its Anthropic Labs division, an AI tool that transforms prompts into polished visual work including designs, interactive prototypes, slide decks, and marketing collateral. The product features conversational prompts and fine-grained editing controls, positioning it as a competitor to established design platforms like Figma. This release marks Anthropic's significant expansion into the creative and visual AI tool market.
Product Launch2026-04-19The VergeOpenAI's Former Sora Boss Is Leaving CompanyBill Peebles, who led OpenAI's Sora video generation team, is leaving the company following OpenAI's decision to shut down the Sora project last month. This departure is part of OpenAI's broader shift away from 'side quests' and consumer-focused moonshots toward enterprise AI priorities. The move signals a strategic pivot as the company consolidates resources around its core products and business model, with several team changes occurring recently.
AI Infrastructure2026-04-19TechCrunch AIAI Chip Startup Cerebras Files for IPOAI chip startup Cerebras has filed for an initial public offering, following recent agreements with Amazon Web Services to use its chips in Amazon data centers and a reported $10+ billion deal with OpenAI. The IPO filing comes as demand for specialized AI hardware continues to grow, with Cerebras positioning itself as a key player in the high-performance computing market for artificial intelligence applications and large language model training.
Product Launch2026-04-19TechCrunch AITesla Expands Robotaxi Service to Dallas and HoustonTesla has expanded its robotaxi service to Dallas and Houston, according to a company social media post. The announcement included a 14-second video showing Tesla vehicles operating without human monitors or drivers. This expansion represents a significant step in Tesla's autonomous vehicle deployment strategy, bringing its robotaxi service to major Texas metropolitan areas and advancing the company's vision for fully autonomous transportation networks.
Open Source2026-04-18WIRED AIMusk v. Altman Trial: Battle for OpenAI's SoulThe upcoming Musk v. Altman trial will determine whether OpenAI has strayed from its founding mission to ensure artificial general intelligence benefits all of humanity. The legal battle represents a fundamental conflict over OpenAI's direction since transitioning to a for-profit structure. The outcome could have significant implications for AI governance and the balance between commercial interests and public benefit missions.
Product Launch2026-04-18The VergeAnthropic's Cybersecurity Model May Mend US Government TiesAnthropic's new cybersecurity model could help repair its strained relationship with the US government after months of public criticism from the Trump administration. The buzzy new security offering might provide common ground despite previous tensions over the company's governance and perceived political leanings. This development highlights how technical capabilities can sometimes transcend political disagreements in the AI sector.
AI Coding2026-04-18TechCrunch AIFactory Hits $1.5B Valuation for Enterprise AI CodingAI coding startup Factory has reached a $1.5 billion valuation after raising $150 million led by Khosla Ventures. The three-year-old company builds AI-powered development tools specifically for enterprise customers, addressing the growing demand for automated coding solutions in large organizations. This funding round underscores the rapid growth and high valuation of specialized AI infrastructure companies.
AI Art2026-04-18TechCrunch AIAnthropic Launches Claude Design for Quick Visual CreationAnthropic has launched Claude Design, a new product that helps non-designers like founders and product managers create quick visuals, prototypes, and marketing collateral through conversational prompts. The tool represents Anthropic's expansion into creative and design workflows, challenging established players like Figma by making visual creation more accessible through natural language interfaces.
AI Coding2026-04-18TechCrunch AICursor in Talks to Raise $2B+ at $50B ValuationAI coding startup Cursor is in advanced talks to raise over $2 billion at a $50 billion valuation, led by returning backers Andreessen Horowitz and Thrive Capital. The massive funding round reflects surging enterprise demand for AI-powered development tools and positions Cursor among the most valuable AI companies globally. This valuation highlights the tremendous investor confidence in AI's transformation of software development.