AI News Summary
Title: AI Weekly: OpenAI unveils Atlas browser, NVIDIA seals $100B OpenAI pact, Meta’s ‘Vibes’ surges, and safety, chips and funding take center stage
Content: OpenAI launched ChatGPT Atlas, an AI-first web browser with built-in ChatGPT, memory, and privacy controls, signaling a direct challenge to Chrome and Safari ahead of multi-platform support. Opera added a Deep Research Agent to Neon that breaks down complex queries and ranks among the top research performers behind Google’s Gemini 2.5 Pro. Microsoft broadened Copilot’s reach across Windows 11—nudging Edge users toward it when visiting ChatGPT or Perplexity—and introduced AI-assisted merge conflict resolution in Visual Studio Code. GitLab rolled out AI agents to automate DevSecOps workflows. Anthropic expanded its Claude platform with Claude Code and modular Claude Skills. Samsung debuted the first Perplexity AI app for smart TVs, Composer launched “Trade With AI” for real-time investing insights, and Autonomize AI unveiled an AI Studio and marketplace featuring 100+ healthcare agents. Alphabet’s Verily introduced Verily Me for easier health data sharing alongside an AI health companion.
Engagement and competition intensified. Meta AI’s mobile app jumped from 775,000 to 2.7 million daily users after launching its TikTok-style “Vibes” feed, though retention and moderation remain open questions. Similarweb data showed ChatGPT’s share of global AI traffic sliding to 74% as Google’s Gemini climbed to 13%, with Perplexity also gaining. Google said Gemini 3.0 will arrive later in 2025 to compete with GPT-5. OpenAI hosted its first international DevDay in São Paulo, underscoring Brazil as its No. 2 user base. Cohere revealed plans for a public offering.
Infrastructure, chips, and energy efficiency dominated headlines. NVIDIA clinched a $100 billion partnership to power OpenAI’s systems, while Oracle unveiled the OCI Zettascale10 supercomputer—800,000 NVIDIA GPUs—backing OpenAI’s Stargate initiative. OpenAI teamed with Hitachi to pursue sustainable, modular, zero-emission AI data centers. Analysts warned that “AI infrastructure debt” is slowing enterprise adoption, and a major AWS outage exposed global dependence on a few cloud providers. On hardware, Cornell researchers reported that GSI Technology’s Gemini-I Associative Processing Unit can match GPU-level performance on large AI workloads while using 98% less energy, potentially reshaping the inference market. Axelera AI launched the Europa chip for real-time, on-device edge AI, and Toyota’s work on AI agents to prevent traffic accidents boosted its stock.
Security, trust, and reliability stayed in the spotlight. A critical flaw (CVE-2025-6515) in Anthropic’s Model Context Protocol allowed session hijacking via predictable IDs, prompting calls to harden session security. Illumio’s AI Insights Agent—already adopted by Microsoft—and GitLab’s agents aim to reduce alert fatigue and accelerate incident response. Streaming platforms faced a surge in AI-powered fraud, and security experts warned that increasingly convincing AI-driven fake apps heighten global cyber risk. New open-source tool SentinelStep improved long-running monitoring agents by optimizing timing and context, pushing task success rates to nearly 40%.
Generative AI ethics and regulation escalated. OpenAI’s Sora 2 drew criticism over unauthorized and fetish deepfakes; after pushback from actor Bryan Cranston and major talent agencies, OpenAI tightened safeguards with stricter opt-in controls for likeness use, energizing support for legislation such as the “NO FAKES” Act. India’s Supreme Court is being pressed to regulate deepfakes and impersonation with national standards and grievance processes. OpenAI called for international collaboration on “sovereign,” democratic AI systems. Aid organizations’ rising use of AI-generated imagery of poverty sparked debates over consent and representation.
Health, science, and research advanced. Qure.ai gained European approval for an AI chest X-ray tool that can screen for tuberculosis in children, including newborns. University of Tokyo researchers paired nanopores with machine learning to classify proteins by electrical signatures, enabling rapid, real-time diagnostics. Verily Me and Autonomize AI aim to streamline health data sharing and workflows. DeepMind’s Demis Hassabis cautioned that AI exhibits “jagged intelligence,” excelling at some complex tasks while failing at simple ones. A new AI tool that could unlock an extra trillion barrels of oil raised alarms over up to 400 billion tonnes of potential additional emissions.
Funding, deals, and enterprise adoption remained brisk. Veeam agreed to acquire Securiti AI for $1.7 billion to fortify data security. LangChain raised $125 million at a $1.25 billion valuation to expand its agent platform. Serval secured $47 million to scale enterprise IT agents; Nexos.ai raised €30 million for secure corporate AI access; Bluwhale collected $10 million to bring AI financial agents to blockchain users (3.6 million globally); Chipmind emerged from stealth with $2.5 million for AI-driven chip design; WideBot raised $3 million and launched an Arabic AI voice agent for WhatsApp; Supermemory secured $2.6 million to commercialize next-gen AI memory; and Zoho rolled out free agentic AI tools for African businesses. The Bank of England is adopting AI to automate tasks and improve forecasting.
Legal and competitive tensions continued. OpenAI moved for summary judgment to dismiss Elon Musk’s lawsuit amid intensifying rivalry with xAI.
News Articles
Title: OpenAI Launches ChatGPT Atlas, a Smart Browser Rival to Chrome and Safari
Content: OpenAI has launched ChatGPT Atlas, a new AI-powered web browser integrating ChatGPT for in-browser research, productivity, and automated tasks. With built-in memory, privacy controls, and multi-platform support coming soon, Atlas aims to redefine web browsing, intensifying competition with Chrome, Safari, and other AI browsers.
Title: GPU-Level AI Performance Achieved With Ultra-Efficient Compute-in-Memory Chip
Content: Cornell researchers report that GSI Technology’s Gemini-I Associative Processing Unit matches GPU performance for large AI workloads while using 98% less energy, potentially disrupting the $100 billion AI inference market and paving the way for greener, faster Edge AI and data center applications.
Title: Meta AI’s ‘Vibes’ Feed Fuels Record Growth, Eyes TikTok’s Audience
Content: Meta AI’s mobile app saw daily users surge from 775,000 to 2.7 million after launching its TikTok-style ‘Vibes’ video feed, putting the app at the forefront of AI-driven mobile engagement as rival platforms experience drops, though long-term retention and moderation remain challenges.
Title: Security Flaw Lets Hackers Hijack Anthropic’s AI Sessions Via Predictable IDs
Content: A critical vulnerability (CVE-2025-6515) in Anthropic’s Model Context Protocol allows attackers to hijack AI chat sessions by predicting session IDs, enabling malicious response injection. Developers are urged to strengthen session security to protect users from hijacks in AI-powered web apps.
Title: Illumio’s AI Insights Agent Aims to End Security Alert Overload
Content: Illumio’s new Insights Agent leverages AI to reduce cybersecurity alert fatigue, offering real-time, role-specific recommendations and instant containment options. Already adopted by Microsoft, the tool helps security teams focus on true threats for faster, smarter incident response.
Title: Serval Raises $47M to Bring Robust AI Agents to Enterprise IT
Content: AI startup Serval landed $47 million in funding to grow its agentic AI platform for IT service management. The system splits automation and execution between specialized agents, giving enterprises like Perplexity granular control, security, and transparency for large-scale AI task management.
Title: Researchers Launch SentinelStep for Smarter, Proactive AI Monitoring Agents
Content: SentinelStep, a new open-source tool, vastly boosts AI agents’ reliability in long-running monitoring tasks—like checking emails or price drops—by dynamically optimizing task timing and context management, pushing success rates up to nearly 40% and enabling more proactive, useful virtual assistants.
Title: Streaming Fraud Surges as Criminals Employ Sophisticated AI Bots
Content: Streaming platforms are battling a wave of fraud powered by AI tools and bots, which manipulate engagement and exploit user accounts. Experts warn that investing in AI-driven security detection is increasingly crucial to protect content, revenue, and users on popular streaming services.
Title: Axelera AI Unveils Europa Chip for Next-Gen Edge AI Applications
Content: Axelera AI has launched the Europa chip, designed to deliver ultra-efficient, real-time AI processing directly on devices like smart sensors and autonomous vehicles, marking a leap forward in high-performance, localized edge computing for businesses and consumers alike.
Title: Autonomize AI Rolls Out Studio and Marketplace to Accelerate Health AI
Content: Autonomize AI’s new AI Studio and healthcare agent marketplace offer over 100 pre-built, domain-specific agents for collaborative workflows among clinicians, business, and IT teams, promising faster deployment and smarter, integrated solutions for pressing healthcare challenges.
Title: Composer’s ‘Trade With AI’ Feature Brings Real-Time Insights to Investors
Content: Investing platform Composer debuts ‘Trade With AI,’ enabling users to harness artificial intelligence for real-time analytics and smarter trading decisions, signaling the growing role of AI in democratizing investment tools for both novice and experienced investors.
Title: GitLab Launches AI Agents to Streamline Security and DevOps Workflows
Content: GitLab’s new AI agents automate routine DevSecOps tasks, filter out unnecessary alerts, and prioritize actionable security insights—helping development teams increase productivity, speed up deployment, and focus on what matters most in the software development pipeline.
Title: Meta AI Surges to 2.7 Million Daily Users After Video Feed Launch
Content: Meta AI’s mobile app has soared to 2.7 million daily users, fueled by the introduction of its ‘Vibes’ feed—showcasing AI-generated short-form videos—as users flock to dynamic content and alternatives to competitors like ChatGPT and Grok.
Title: Cohere Plans IPO, Marking Milestone for AI Model Creators
Content: Cohere, a leading competitor to OpenAI, has revealed plans for a public offering—potentially the first major IPO for a dedicated AI model company—as demand for pure AI investments grows among global investors.
Title: Samsung Debuts Perplexity AI App, Making TVs Smarter Than Ever
Content: Samsung has launched the world’s first Perplexity AI-powered app for smart TVs, turning them into interactive assistants that answer questions on everything from movies to travel—redefining entertainment and convenience for millions.
Title: Veeam Acquires Securiti AI in $1.7 Billion Data Security Deal
Content: Veeam Software is acquiring Securiti AI for $1.7 billion to boost its data management and security capabilities, underscoring the rising importance of AI-driven data protection across global markets.
Title: OpenAI Faces Scrutiny Over Sora’s Deepfake and Fetish Content
Content: OpenAI’s Sora video generator has sparked controversy over unauthorized and fetish deepfake videos, prompting public outcry and calls—including from actor Bryan Cranston and major talent agencies—for stronger protections of performers’ rights and privacy.
Title: Anthropic’s Claude Platform Expands With Coding and Custom Skills
Content: Anthropic has launched Claude Code, an AI-powered coding assistant, and introduced Claude Skills—modular tools that help large language models perform specialized tasks—making advanced coding and workflow automation more accessible to users.
Title: Aid Agencies Embrace AI-Generated Images, Sparking Ethical Debate
Content: Humanitarian organizations and stock platforms are increasingly using AI-generated images of poverty for campaigns and storytelling, raising tough questions about consent, representation, and the ethics of digital imagery in global media.
Title: Qure.ai Launches First AI TB Screening Tool for Youngest Children
Content: Indian healthtech startup Qure.ai has secured European regulatory approval for an AI-powered chest X-ray tool that can now screen children as young as newborns for tuberculosis, aiming to dramatically improve early detection and outcomes.
Title: Microsoft Pushes Copilot AI Against ChatGPT and Rivals in Windows
Content: Microsoft is integrating and promoting its Copilot AI assistant throughout Windows 11, actively encouraging users to try it when accessing platforms like ChatGPT and Perplexity in Edge, intensifying the competition among AI assistants.
Title: Alphabet’s Verily Launches AI App for Seamless Health Data Sharing
Content: Alphabet’s Verily has unveiled Verily Me, an AI-powered app that makes sharing health records easier across providers while debuting an AI health companion—highlighting the rapid evolution of digital health technology and data accessibility.
Title: ChatGPT’s Market Share Slips as Google Gemini Surges Worldwide
Content: ChatGPT’s share of global AI traffic dropped to 74% as Google’s Gemini climbed to 13%, according to Similarweb’s latest data. The shift highlights growing global competition in generative AI, with emerging players like Perplexity also gaining ground.
Title: OpenAI and Hitachi Partner to Build Sustainable, Next-Gen AI Data Centers
Content: OpenAI and Hitachi have joined forces to develop advanced AI infrastructure, aiming for sustainable, zero-emission data centers and accelerated modular construction. This strategic alliance positions both companies at the forefront of efficient, scalable, and eco-friendly data management.
Title: Google to Release Gemini 3.0 This Year, Challenging GPT-5
Content: Google CEO Sundar Pichai announced the Gemini 3.0 AI model will launch later in 2025, delivering smarter, faster AI agents and positioning Google to compete directly with GPT-5 and other leading models.
Title: NVIDIA Clinches $100B OpenAI Deal, Tightening Grip on AI Chips
Content: NVIDIA secured a $100 billion partnership with OpenAI, ensuring its AI chips will power OpenAI’s systems and preventing rival Google from gaining ground. The deal cements NVIDIA’s dominance as critical AI infrastructure becomes increasingly competitive.
Title: Courts Pressed to Regulate AI as Deepfake Threats Mount
Content: India’s Supreme Court is being urged to introduce a robust AI regulatory framework, especially targeting deepfakes and impersonation. The petition demands national standards and grievance mechanisms from tech giants to protect privacy and fundamental rights.
Title: OpenAI Holds First International DevDay in Brazil, Spotlights Nation’s AI Growth
Content: OpenAI’s inaugural DevDay Exchange in São Paulo celebrated Brazil’s rising AI influence, gathering top developers and underscoring the country’s global significance as the second-largest user of OpenAI tools.
Title: Chipmind Raises $2.5M to Revolutionize Chip Design with AI Agents
Content: European startup Chipmind has emerged from stealth with $2.5 million in funding, unveiling AI agents that streamline chip design and manufacturing, promising to accelerate semiconductor innovation and meet soaring demand.
Title: AI Infrastructure “Debt” Slows Adoption, Warns Tech Analysts
Content: Experts caution that underinvestment in foundational AI infrastructure is hampering the swift rollout of new AI technologies, urging businesses to prioritize robust data systems and integration to unlock AI’s full potential.
Title: Zoho Launches Free AI Tools to Power African Businesses
Content: Zoho unveiled advanced, agentic AI capabilities—available at no cost—to automate tasks across collaboration, customer service, and HR for African firms, marking a major push for accessible enterprise AI.
Title: Bluwhale Secures $10M to Deploy AI Agents on Blockchain
Content: Decentralized AI network Bluwhale raised $10 million in Series A funding to roll out AI-powered financial agents on blockchain platforms, aiming to transform personalized digital asset services for over 3.6 million users worldwide.
Title: WideBot Debuts Arabic AI Voice Assistant for WhatsApp Business
Content: WideBot, an Egyptian AI startup backed by $3 million in new funding, has launched a pioneering Arabic-language AI voice agent for WhatsApp, enabling real-time business communications across diverse Middle Eastern dialects.
Title: Supermemory Startup Secures $2.6M to Boost AI’s Recall Intelligence
Content: Teenage founder Dhravya Shah’s startup Supermemory raised $2.6 million to commercialize next-generation AI memory technology, enabling smarter, context-aware systems now attracting top Silicon Valley clients.
Title: Bank of England Adopts AI to Modernize Internal Operations
Content: The Bank of England is integrating AI tools to automate tasks, analyze big data, and improve forecasting, strengthening its position as a digital innovator in the financial sector.
Title: OpenAI Tightens Sora 2 Controls After Celebrity Likeness Controversy
Content: After unauthorized use of Bryan Cranston’s likeness through its Sora 2 video AI, OpenAI has pledged to bolster protections against unlicensed use of celebrity IP—responding to mounting industry and union pressures.
Title: Nexos.ai Lands €30M to Accelerate Secure AI Use in Enterprises
Content: Lithuanian startup Nexos.ai raised €30 million to streamline safe AI adoption in corporations, helping manage sensitive data and comply with regulations while providing secure access to hundreds of AI tools.
Title: Toyota Shares Jump on AI Project Tackling Traffic Accidents
Content: Toyota’s stock surged following news that the automaker is developing advanced AI agents aimed at preventing traffic accidents by analyzing vehicle and infrastructure data, underscoring its tech-forward safety ambitions.
Title: Hollywood Spurs OpenAI to Strengthen Safeguards on AI-Generated Likenesses
Content: Following intense backlash from Hollywood studios and talent agencies, OpenAI has introduced stricter opt-in policies and new safeguards for its Sora 2 video tool to prevent unauthorized use of celebrity likenesses, fueling calls for legislation like the “NO FAKES” Act and setting a precedent in the ongoing debate over performers’ rights and AI-generated content.
Title: Oracle Debuts World’s Largest AI Supercomputer Backing OpenAI’s Stargate Initiative
Content: Oracle has unveiled the OCI Zettascale10 supercomputer, equipped with 800,000 NVIDIA GPUs and powering OpenAI’s ambitious Stargate project, signaling a bold advance in global AI infrastructure despite questions over its real-world performance and industry competition.
Title: New AI Tool Risks Unleashing Trillion Barrels of Oil, Stoking Climate Fears
Content: A cutting-edge AI system could unlock an extra trillion barrels of oil, but scientists warn this breakthrough could result in up to 400 billion tonnes of additional climate pollution, intensifying the conflict between technological progress and urgent environmental action.
Title: AWS Outage Exposes Fragility of Global Internet and Critical AI Services
Content: A major Amazon Web Services glitch caused widespread disruptions to banks, apps, and AI platforms, highlighting the world’s reliance on a handful of cloud providers and raising alarms about the vulnerability of essential digital infrastructure.
Title: Google DeepMind CEO Highlights AI’s “Jagged Intelligence” Limitations
Content: DeepMind CEO Demis Hassabis warns that despite remarkable advances, today’s AI excels at complex tasks but can still stumble on simple ones, stressing that true human-level reasoning remains out of reach and demanding new research into AI’s core capabilities.
Title: OpenAI Pushes for Global Collaboration on Sovereign, Democratic AI
Content: OpenAI’s Chris Lehane urges international cooperation for the responsible development of “sovereign” national AI systems, emphasizing democratic values and warning that the rapid rise of superintelligent AI will transform societies and economies worldwide.
Title: AI-Driven Fake Apps Outpace Security, Raising Global Cyber Risks
Content: Cybercriminals are harnessing AI to create highly convincing fake apps, making them increasingly difficult for users to detect and prompting security experts to call for greater vigilance, updated safety measures, and widespread education to protect against costly digital scams.
Title: LangChain Raises $125M to Expand User-Friendly AI Agent Platform
Content: Open-source platform LangChain has secured $125 million in fresh funding, reaching a $1.25 billion valuation as it expands developer-friendly tools that simplify building and tracking advanced AI agents across multiple models.
Title: Molecular Profiling Revolutionized by Nanopore-AI Fusion
Content: University of Tokyo researchers have developed a breakthrough technique pairing nanopore technology and machine learning to rapidly classify proteins by their electrical signatures, opening new frontiers in real-time biomedical diagnostics and disease detection.
Title: Opera Neon Adds Parallel AI Agent, Challenging Google’s Gemini in Research Tasks
Content: Opera’s browser now features the Deep Research Agent, breaking down research questions for faster, more thorough answers and scoring as one of the best AI research performers behind Google’s Gemini 2.5 Pro.
Title: Visual Studio Code Launches AI-Powered Merge Conflict Resolution
Content: Microsoft’s Visual Studio Code now offers AI-assisted tools to help developers quickly resolve merge conflicts, streamlining collaboration and making coding projects more efficient for users across all major platforms.
Title: Elon Musk’s Lawsuit Against OpenAI Faces Push for Summary Dismissal
Content: OpenAI has filed for summary judgment in Elon Musk’s lawsuit against the company, accusing Musk—whose own startup xAI competes in the field—of using legal action as a competitive tactic amid heightened rivalry in the AI sector.
