AI News Today: OpenAI Launches ‘Operator’ Beta, Apple Intelligence 2.0 Leaked, & Nvidia Stumbles

Ten days after the artificial intelligence industry was shaken by massive releases from Google and Anthropic, the silence from San Francisco has finally broken. Today’s AI news is dominated by OpenAI’s counter-attack: the launch of the limited beta for “Operator,” a browser-native autonomous agent designed to execute complex tasks on the web. However, the news cycle is far from a victory lap for OpenAI; fresh leaks regarding Apple Intelligence 2.0 suggest the iPhone maker is preparing to reduce its reliance on ChatGPT with its own “Ajax” model, while hardware giant Nvidia (NVDA) faces a rare stock stumble following supply chain rumors regarding its next-generation Rubin chips.

OpenAI Strikes Back: ‘Operator’ Beta Enters the Agent Wars

For the past week, the narrative has centered on Google’s Antigravity platform and Anthropic’s Claude Opus 4.5, both of which introduced powerful “agentic” capabilities—the ability for AI to use computers rather than just chat. Today, OpenAI officially entered this new battlefield with the release of Operator.

Unlike Anthropic’s “Computer Use,” which controls the entire desktop environment, Operator is a browser-native agent. It lives within the browser as an extension, designed specifically to navigate websites, fill out forms, and execute multi-step workflows. In the official press release, OpenAI demonstrated Operator booking a multi-city flight itinerary, researching and compiling a spreadsheet of holiday gifts, and even debugging code directly within a cloud-based IDE.

Crucially, OpenAI is emphasizing safety to distinguish itself from Google’s more open-ended approach. The beta includes a prominent “Red Button” feature—a human-in-the-loop oversight mechanism that requires user confirmation before the agent executes any financial transaction or sends external communications. This appears to be a direct response to the growing regulatory scrutiny over autonomous AI, including the recent youth protection laws introduced in Virginia earlier this month.

Operator is currently available only to a small subset of “Plus” and “Pro” users in the United States, with a wider rollout planned for early 2026. For OpenAI, this release is a critical survival move. With enterprise clients like Salesforce already defecting to Google’s ecosystem, OpenAI needed to prove that it could offer more than just conversation. Operator is that proof, shifting the company’s value proposition from “generating text” to “taking action.”

Apple Intelligence 2.0 Leaks: The Rise of the ‘Ajax’ Model

While OpenAI fights to retain its users, its most significant consumer partner appears to be plotting a breakup. Supply chain sources and code discovered in the iOS 19.3 beta have shed light on the roadmap for Apple Intelligence 2.0, slated for a Spring 2026 release.

The leaked code references “Ajax,” Apple’s internal proprietary Large Language Model (LLM). According to insiders, the iPhone 17 Pro will feature an optimized, 30-billion parameter version of Ajax running entirely on-device. This development is significant because it powers a new “Conversational Siri” capable of maintaining long-term context across different apps without pinging the cloud.

Currently, Apple Intelligence relies on a partnership with OpenAI to handle complex queries via ChatGPT. The leaks suggest that Apple Intelligence 2.0 will relegate ChatGPT to a fallback option for only the most niche requests, with the native Ajax model handling the vast majority of reasoning and generation tasks. If accurate, this shift could drastically reduce the traffic OpenAI receives from Apple devices, potentially impacting their user growth metrics just as they attempt to monetize through the controversial ad strategy leaked earlier this month.

Nvidia Stock Stumbles Amidst Rubin GPU Delay Rumors

In the hardware sector, the unstoppable momentum of Nvidia hit a speed bump today. NVDA stock is down approximately 4-5% in pre-market and early trading following a report from manufacturing partners in Taiwan.

The report alleges that yields for the upcoming Rubin (R100) architecture—the successor to the wildly popular Blackwell series—are lower than anticipated due to complexities in the packaging process. While Nvidia has not officially commented, the market’s reaction highlights the extreme volatility and high expectations surrounding the AI supply chain. Investors are jittery; any sign that the “picks and shovels” of the AI gold rush might be delayed causes immediate ripples.

This stumble has created a momentary opening for competitors. AMD saw a slight uptick in trading as investors looked to hedge their bets, and analysts are paying renewed attention to Google’s custom silicon strategy. With Gemini 3 proving the efficacy of Google’s TPU v6 pods, the reliance on Nvidia’s timeline may be slightly less absolute for the hyperscalers than it was a year ago.

FTC Launches Probe into Generative AI Ad Injection

Regulatory pressure continues to mount following the December 1 revelations that OpenAI was testing search ads in ChatGPT. Today, the Federal Trade Commission (FTC) confirmed it has opened a preliminary inquiry into “Generative AI Ad Injection.”

The probe is not targeting OpenAI exclusively but is looking at the broader industry practice of “blended content,” where AI-generated answers are indistinguishable from paid sponsorships. FTC Chair Lina Khan stated in a briefing that users have a right to know if the advice they receive from an AI—whether it’s a product recommendation or a travel itinerary—is organic or paid for. This investigation could force companies like OpenAI, Perplexity, and Google to redesign their interfaces to clearly label sponsored outputs, potentially complicating their monetization roadmaps.

Rapid Updates: Midjourney Video and Tesla Optimus

In other news, the creative AI sector remains competitive. Rumors indicate that Midjourney is preparing to launch Midjourney v7 next week, which will reportedly feature a “Video Mode” to compete directly with Google’s newly released Nano Banana Pro.

Meanwhile, Elon Musk took to X (formerly Twitter) to share a new video of the Tesla Optimus robot autonomously folding laundry. While Musk claims a “2026 household release,” experts remain skeptical of the timeline, though the dexterity demonstrated in the video shows significant improvement in fine motor control compared to previous demonstrations.

The Battle for Autonomous Execution Heats Up

As we approach the end of 2025, the industry focus has decisively shifted from “chatbots” to “agents.” The release of OpenAI’s Operator beta confirms that the next frontier is not about who has the smartest model, but whose AI can most effectively navigate the human web to get things done. However, with Apple moving toward independence and Nvidia facing supply constraints, the path forward is complex. The winners of 2026 will likely be the companies that can balance this new autonomous power with the growing demands for privacy, safety, and regulatory compliance.

• What Is an AI Agent for Email Marketing? The Complete 2026 Guide
• AI Agents for Task Automation: How They Work, Benefits, and Top Use Cases
• What’s the difference between an AI Agent and an LLM?
• AI Agents for Marketing: How Autonomous Intelligence Is Transforming Digital Campaigns
• 5 best AI agent builder platforms in 2025