The AI race shows no signs of slowing down. In the past week, three major developments hit the tech news cycle: a leaked Google Gemini Omni video model, an eye-opening stat from Airbnb about AI-generated code, and the official launch of WhatsApp Business AI tools. Here is what you need to know, with context on why each matters.

Google Gemini Omni video model leaks online

Google’s next-generation AI model, Gemini Omni, appears to have surfaced early. According to reports circulating this week, a video demonstration of the model leaked on social media, showing what is believed to be a multimodal system capable of processing video input in real time.

The leaked footage, which has not been officially confirmed by Google, reportedly depicts Gemini Omni analyzing a live video stream and responding with spoken commentary — a capability that exceeds what current public versions of Gemini offer. If authentic, this would place Google in direct competition with OpenAI’s GPT-4o and other “omni” models that combine vision, audio, and text understanding in a single neural network.

What makes Gemini Omni notable is the emphasis on video. While Google already offers Gemini 1.5 Pro with video understanding (it can watch uploaded video and answer questions), Omni appears to be designed for instant, real-time processing — more like a human watching a scene unfold than a model reviewing a file after the fact. That shift from batch to live interaction is significant. It opens the door to applications like real-time tutoring, live event commentary, on-the-fly translation of video content, and accessibility tools for people with visual impairments.

Google has not commented on the leak. The company typically does not discuss unreleased products, and it is possible that the video is from an internal prototype or a demo meant for a partners-only event. Still, the appearance of such a demonstration suggests that Google is pushing hard to close the gap with OpenAI, which has set the pace in multimodal AI since the launch of GPT-4 with vision in late 2023.

For now, developers and AI watchers will have to wait for an official announcement. But the leak confirms one thing: the next big battlefield in AI is real-time video understanding, and Google is determined to play there.

Airbnb says AI now writes 60% of its code

This week, Airbnb made a surprising disclosure. The company announced that AI tools — likely large language models integrated into its development pipeline — now generate approximately 60% of its production code. The claim came during a public presentation or interview, though the exact venue and speaker have not been confirmed in the source material.

If accurate, that figure is remarkable. It suggests that the majority of code shipped by one of the world’s most popular travel platforms is written not by human engineers, but by AI models that translate natural language prompts or structured specifications into working software.

Airbnb is not the first company to talk about AI-assisted coding. GitHub Copilot, Amazon CodeWhisperer, and Google’s own coding tools have been widely adopted across the industry. But few companies have publicly quantified how much of their codebase is AI-generated. Airbnb is essentially saying that two out of every three lines of code they deploy come from a machine, with humans presumably reviewing, editing, and integrating the output.

This raises obvious questions about code quality, security, and long-term maintainability. AI-generated code can introduce subtle bugs or security vulnerabilities that human reviewers might miss, especially when the AI is working with complex or domain-specific logic. Airbnb has not released details about its review process, but the company’s willingness to share such a high percentage indicates confidence in its safeguards.

For the broader tech industry, the Airbnb stat is a signal. If a major consumer-facing marketplace can safely rely on AI for 60% of its code, then the barrier to adopting AI coding tools for other companies is lower than many expected. It also puts pressure on competitors — if your engineers are not using AI to generate code at scale, you may be leaving productivity gains on the table.

WhatsApp Business launches AI-powered features

Meta is expanding its AI push to the business messaging space. This week, the company announced the launch of AI tools for WhatsApp Business, a platform used by millions of small and medium enterprises worldwide to communicate with customers.

The new features appear to include AI-generated replies, message drafting assistance, and perhaps smart categorization of customer inquiries. These capabilities are designed to help small businesses handle customer service more efficiently without requiring a dedicated support team. A shop owner who gets dozens of messages a day can use AI to automatically suggest responses to common questions about hours, pricing, or order status, freeing up time for more complex interactions.

WhatsApp Business already offers a catalog feature, quick replies, and automated away messages. The addition of AI takes that a step further by making the automation smarter and more context-aware. Instead of typing out a boilerplate reply, a business owner can let the AI draft a response that fits the specific query, then either send it as-is or edit it before hitting send.

Meta has been slowly integrating its in-house AI, known as Meta AI, across its family of apps, including Facebook, Instagram, and Messenger. WhatsApp has so far been the most cautious — consumer WhatsApp received AI search and chat features only recently, and the company has been deliberate about not overwhelming users with AI. The WhatsApp Business launch represents a natural next step: small businesses are often the most eager adopters of tools that save time and money, and AI fits that need well.

The timing also lines up with a broader trend. Google and Microsoft have been pushing AI into their business messaging and collaboration products for months. With this launch, Meta is making sure it does not fall behind in the enterprise communication space, even if WhatsApp’s core business remains consumer messaging.

Baidu’s Ernie 5.1: a quick mention

The source briefing also mentions Baidu’s Ernie 5.1, though no specific details are available beyond the name. Given the headline’s focus on Gemini, Airbnb, and WhatsApp, the Baidu update likely rounds out a week of AI announcements from China. Ernie is Baidu’s answer to ChatGPT, and version 5.1 presumably brings improvements in reasoning or multimodal capability. Without concrete information, we note it as a reminder that the AI race is truly global — while the U.S. big tech companies dominate headlines, Chinese firms like Baidu continue to iterate at a rapid pace.

What it all means

Three stories from one week: a leaked Google video model, a stunning AI code generation stat from Airbnb, and a pragmatic AI tool launch from WhatsApp Business. Each touches a different part of the AI ecosystem — research, engineering productivity, and small business application.

The common thread is that AI is moving from experimental to operational. Google wants AI to see and understand the world in real time. Airbnb treats AI as a core part of its software engineering workflow. Meta is putting AI directly into the hands of shop owners who need practical help today.

For anyone watching the space, the signal is clear: the next phase of the AI boom is not about demos or benchmarks. It is about integration at scale. The companies that figure out how to deploy AI safely and effectively — whether in code, customer service, or video analysis — will be the ones that define the decade.

The week’s news also shows how quickly the conversation shifts. Just a few months ago, text-based chatbots were the headline. Now the focus is on video, code generation, and business tools. The pace is accelerating, and staying informed requires tracking not just the biggest names but also the concrete ways AI is being used behind the scenes.

SysCall News will continue to monitor these developments as more details emerge on Gemini Omni, Airbnb’s coding practices, and WhatsApp Business’s AI rollout. For now, the message is simple: AI is not coming — it is already here, writing code, answering customers, and watching the world in real time.