🗞 OpenAI's Web Assistant Launch, $100B AI Infrastructure Plan, and Trump's AI Order

🔎 The Latest on the AI Frontier:

• OpenAI Unveils Operator: Autonomous Web Assistant for Pro Users

• OpenAI and SoftBank Lead $100B Stargate Project for AI Infrastructure

• Trump Issues Executive Order Prioritizing AI Innovation

• Reliance Plans 3GW Data Centre in India with Nvidia Partnership

• OpenAI's Operator Tool Extends Data Retention to 90 Days

• Anthropic Launches Citations API to Enhance AI Accuracy

• Meta AI Chief Predicts Major Architecture Shift by 2028

• Other news you might find interesting

🤖 OpenAI launches Operator, an AI agent that can autonomously perform web tasks for Pro users

• Operator uses its own browser to handle tasks like form filling and online shopping, powered by a new Computer-Using Agent (CUA) model that sets benchmarks in WebArena and WebVoyager tests.

• Built-in safety features include mandatory user approval for significant actions, takeover mode for sensitive information, and defenses against malicious websites.

• Initially available to U.S. Pro users only, with partnerships including DoorDash, Instacart, and Uber, and plans to expand to Plus, Team, and Enterprise users.

💰 OpenAI and SoftBank to lead $100B Stargate Project with equal $19B investments

• The AI-focused venture brings together major tech players including Oracle and MGX, with OpenAI and SoftBank each taking approximately 40% stake in the initial $45B commitment.

• Project aims to build U.S. data centers starting in Texas, with potential to reach $500B investment over four years for AI applications including mRNA vaccines.

• Notable concerns raised by industry leaders - Anthropic's CEO and Elon Musk question the financing, with Musk claiming SoftBank has "well under" $10B secured.

🎯 Trump signs AI order prioritizing innovation over regulation

• New order requires 180-day AI action plan focused on eliminating "ideological bias," with venture capitalist David Sacks appointed as Special Advisor for AI and Crypto.

• Directs review of all Biden-era AI policies, particularly those requiring companies to demonstrate their AI tools don't harm the public.

• Critics argue the order's focus on removing "barriers to innovation" could unwind existing protections against AI discrimination in areas like medical diagnosis and facial recognition.

💼 Reliance plans world's largest $30bn data centre in India with 3GW capacity

• India's Reliance conglomerate aims to construct a multi-gigawatt facility in Jamnagar, powered entirely by green energy through solar, wind, and hydrogen projects.

• The facility will house Nvidia's next-generation GB200 chips, building on the partnership announced at Nvidia's AI Summit in Mumbai last October.

• Project joins competitive landscape of mega data centers, including Meta's Louisiana facility, Tract's Arizona project, and Kevin O'Leary's proposed Alberta site.

🤖 OpenAI extends data retention period for new AI agent Operator to 90 days after deletion

• OpenAI's new Operator tool, available to $200/month ChatGPT Pro subscribers, can automate tasks like booking travel and making restaurant reservations with a built-in browser capability.

• The company will store deleted chats and screenshots for up to 90 days - triple the retention period of ChatGPT's 30-day policy - citing the need to monitor potential abuse of this new technology.

• Data may be accessed by authorized OpenAI personnel and trusted service providers for abuse investigation and legal matters, even after user deletion.

🚀 Anthropic launches Citations feature to boost AI response verification and trust

• The new Citations API feature enables Claude to automatically reference specific sentences from source documents, improving recall accuracy by up to 15% compared to prompt-based approaches.

• Available on Anthropic API and Google Cloud's Vertex AI, Citations uses standard token-based pricing without charging for quoted text in outputs.

• Thomson Reuters and Endex report significant improvements - with Endex noting a reduction in source hallucinations from 10% to 0% and 20% more references per response.

🔮 Meta's AI chief forecasts major shift in AI landscape by 2028

• LeCun predicts current LLM paradigm will be obsolete within 3-5 years, replaced by new AI architectures with improved world understanding, memory, reasoning, and planning capabilities.

• Meta is developing "world models" to give AI systems common sense and better understanding of physical reality, moving beyond pattern recognition.

• Next decade could be "decade of robotics," with AI systems achieving cat-like understanding of the physical world through combined advances in AI and robotics.

🖨 More news you might find interesting:

• Musk-Altman feud escalates over Trump's $500B AI project

• Internal documents reveal Google Cloud division worked directly with IDF on AI tools despite public denials

• Perplexity brings AI assistant to Android in 15 languages, capable of multi-app actions like ride-hailing and calendar management.

• Galileo launches AI agent testing platform with $68M in funding to prevent deployment errors

• Samsung integrates AI into SmartThings platform to automate home routines based on user activities

• Mark Lemley, one of the most-cited legal scholars and Meta's key intellectual property lawyer, has resigned from his position at the company.

• Trump mandates new working group to develop crypto regulations, protect banking services, and explore creating national cryptocurrency reserve.

• AI model shows promise in early colorectal cancer detection with 93.8% sensitivity

Have any feedback? Send us an email