google, openai and nvidia Updates

Google's DeepMind has unveiled Gemini 2.5 Computer Use, an AI model designed to interact with websites and applications much like a human. This model can navigate the web, fill out forms, and perform actions like clicking and typing based on text commands. Available to developers via the Gemini API, it has shown strong performance in benchmarks, even outperforming competitors like Claude and OpenAI in web and mobile control. This advancement is further supported by 1Password's new Secure Agentic Autofill feature, which provides secure credential access to AI agents operating in browsers, authorizing logins only after human approval and without exposing sensitive information. Meanwhile, the broader AI industry sees continued investment, with Nvidia CEO Jensen Huang downplaying concerns of an AI bubble, citing strong company profits and financial health as drivers, unlike the dot-com era. However, the market is not without its fluctuations, as seen in Oracle's stock dip due to server rental profit margins and Tesla's share decline. On the geopolitical front, AI presents national security risks, with OpenAI reporting state-sponsored adversaries using AI for cyberattacks and influence operations. In expansion news, AI startup Anthropic plans to open its first India office in Bengaluru in early 2026, recognizing India as its second-largest market for its chatbot Claude, and faces competition from Google's Gemini and OpenAI, which has also established a presence in India. In other AI applications, Finland is using drones and AI to optimize traffic flow in Helsinki's West Harbour, while defense tech firms like VisionWave and PVML are partnering to develop secure, real-time intelligence systems for autonomous AI agents, potentially reducing development cycles by up to 70%.

Key Takeaways

• Google's Gemini 2.5 Computer Use model allows AI agents to browse the web, fill forms, and interact with websites like humans.
• 1Password's Secure Agentic Autofill provides secure credential access for AI agents, requiring human authorization for website logins.
• Nvidia CEO Jensen Huang believes the current AI investment surge is driven by company profits and financial health, not a speculative bubble.
• Oracle's stock experienced a drop due to concerns over profit margins on Nvidia-powered servers, impacting AI infrastructure stocks.
• Tesla shares also declined following reports of price reductions on its Model Y vehicles.
• OpenAI reports that state actors from China, Russia, North Korea, and Iran are using AI for cyberattacks, espionage, and influence operations.
• AI startup Anthropic will open its first India office in Bengaluru in early 2026, with India being its second-largest market.
• Helsinki, Finland, is employing drones and AI to manage traffic flow at its West Harbour.
• VisionWave and PVML are collaborating to create secure AI systems for defense, aiming to reduce development cycles by up to 70%.
• Zelda Williams has condemned the creation of unauthorized AI-generated videos featuring her late father, Robin Williams.

Google's Gemini AI can now browse web, fill forms

Google's DeepMind AI lab has created a new version of its Gemini 2.5 Pro model called Gemini 2.5 Computer Use. This AI can use a virtual browser to surf the web, find information, fill out forms, and take actions on websites based on simple text commands. While not directly available to consumers, it can be accessed through platforms like Browserbase and Google's own developer tools. Early tests show it can navigate websites, summarize content, and even pass CAPTCHAs, though it does not currently create or edit files directly.

Google's Gemini 2.5 model helps AI agents use websites

Google has launched the Gemini 2.5 Computer Use model, an updated version of its Gemini 2.5 Pro system. This new model allows AI agents to directly interact with websites and applications by performing actions like clicking, typing, and scrolling. It works by processing user requests, screenshots, and action history to generate necessary actions, repeating a loop until the task is complete. The model is available through the Gemini API in Google AI Studio and Vertex AI, and Google's payments team has already used it to improve software testing efficiency.

Google's Gemini 2.5 AI navigates web like a human

Google's DeepMind has released Gemini 2.5 Computer Use, a new AI model that can navigate the web and interact with websites like a human. It analyzes user requests, screenshots, and past actions to perform tasks such as clicking, typing, and filling forms. This model builds on Gemini 2.5 Pro and is available to developers through Google's AI platforms. While it excels in browser interactions, it currently lacks direct operating system control, unlike some competitors like OpenAI and Anthropic.

Gemini 2.5 Computer Use model previewed with strong web performance

Google is offering developers a preview of its Gemini 2.5 Computer Use model, designed to interact with graphical user interfaces like browsers and websites. The model operates in a loop, taking user requests, screenshots, and action history to perform tasks such as clicking, typing, and scrolling. It shows strong performance in web and mobile control benchmarks, outperforming competitors like Claude and OpenAI. Gemini 2.5 Computer Use is available through the Gemini API in Google AI Studio and Vertex AI.

1Password secures AI agents with new autofill integration

1Password has launched Secure Agentic Autofill, a new feature for its Enterprise Password Manager that securely provides credentials to AI agents operating in browsers. Available through an early access integration with Browserbase, this feature allows AI agents to authenticate with websites without exposing sensitive information. It includes human-in-the-loop authorization, privacy protections, and just-in-time authentication, ensuring credentials remain secure. This integration aims to allow organizations to scale AI confidently by providing a trusted layer for credential access.

1Password secures AI browser agents with new autofill tool

1Password has introduced Secure Agentic Autofill, a new feature designed to enhance login security for AI agents browsing the web. This tool allows AI bots to access necessary credentials by injecting them directly into the browser only after a human approves the request. The AI agent never sees the actual credentials, as 1Password handles the secure transfer. Initially available in early access through Browserbase, this feature aims to prevent security breaches caused by AI bots remembering sensitive information.

Nvidia CEO downplays AI bubble concerns

Nvidia CEO Jensen Huang recently addressed concerns about a potential Artificial Intelligence (AI) bubble. In an interview with CNBC, Huang suggested that the current surge in AI investment and company valuations is different from the dot-com bubble of 2000. He highlighted key distinctions that he believes make the current situation more sustainable.

Stocks react to Oracle AI fears; Tesla shares drop

The stock market saw a modest pullback as investors reacted to news that Oracle is experiencing low profit margins from renting servers equipped with Nvidia chips. This led to a drop in Oracle's stock and other AI infrastructure companies. Meanwhile, Tesla shares also declined following reports of cheaper Model Y vehicles.

AI and technology create national security risks

Artificial intelligence combined with widespread technology presents a growing national security risk. OpenAI reported that adversaries like China, Russia, North Korea, and Iran are using AI tools for cyberattacks, espionage, and influence operations. These AI-powered campaigns can scale deception and manipulate digital environments quickly. Examples include fake personas on social media, AI-assisted malware development, and AI-generated fake resumes to infiltrate corporate systems. This trend requires cleared professionals to verify information beyond resumes, rethink endpoint security, and report unusual digital behavior.

AI startup Anthropic to open India office in 2026

Artificial intelligence startup Anthropic plans to open its first office in India in early 2026, located in Bengaluru. This move aims to capitalize on India's increasing adoption of AI tools, driven by enterprise spending and a skilled workforce. India is already Anthropic's second-largest market for its chatbot Claude. The company faces competition from other AI players like Google's Gemini and OpenAI, which has also registered as a legal entity in India. Anthropic recently announced plans to triple its workforce to meet growing demand.

Robin Williams' daughter condemns AI videos of him

Zelda Williams, daughter of the late actor Robin Williams, has strongly criticized AI-generated videos that recreate her father. She described these deepfake videos as 'disgusting' and expressed her opposition to unauthorized digital recreations of deceased individuals. The comments came after an AI video of Robin Williams began circulating online.

Finland uses drones and AI to manage traffic

Helsinki, Finland is using drones and artificial intelligence (AI) to improve traffic flow at its busy West Harbour. Since 2023, drones have collected data on passenger routes and traffic patterns to help optimize connections between ferries, trams, metro, and trains. The data is processed by AI to create simulations and test traffic management strategies virtually. While efforts are made to comply with privacy laws by anonymizing data, some experts raise concerns about potential privacy violations and cybersecurity risks.

VisionWave and PVML partner for secure AI in defense

Defense technology company VisionWave and secure data-AI infrastructure firm PVML have formed a strategic collaboration. This partnership aims to create secure, real-time intelligence systems for mission-critical operations using live data without compromising security. The integration combines VisionWave's radar and computer vision systems with PVML's infrastructure, enabling autonomous AI agents to plan and execute missions dynamically. Early projects suggest development cycles could be reduced by up to 70%.

AI stock rally driven by profits, not superintelligence

A new Goldman Sachs report suggests the current surge in AI stock valuations is driven by strong profits and solid company finances, not by speculation about future superintelligence. Companies like Nvidia and Microsoft show significant profit growth and strong balance sheets, unlike during the dot-com bubble. While a correction is possible, the rally is currently based on real performance and productivity gains rather than a belief in imminent human-level AI.

Sources

• Google's AI can now surf the web for you, click on buttons, and fill out forms with Gemini 2.5 Computer Use
• Google (GOOGL) Launches New Gemini 2.5 Model to Help AI Agents Use Websites
• Google's Gemini 2.5 Computer Use model can navigate the web like a human
• ‘Gemini 2.5 Computer Use’ model enters preview with strong web, Android performance
• 1Password tackles AI credential risks with new Agentic Autofill integration for Browserbase
• 1Password says it can fix login security for AI browser agents
• Nvidia Chief Huang downplays AI bubble concerns in CNBC interview (NVDA:NASDAQ)
• Stocks Weather Oracle Fears; Tesla Skids On Cheap Model Y
• AI + Ubiquitous Tech = National Security Risk Multiplier
• Anthropic to open India office in 2026
• Robin Williams’ daughter rails against AI videos of him
• Finland is using drones and AI to tackle congestion. Is it safe?
• VisionWave Announces Strategic Collaboration with PVML to Advance Secure, Real-Time AI for Mission-Critical Operations
• AI Stocks Fly High—but Not on Dreams of Superintelligence