OpenAI's TikTok for AI content and ChatGPT Pulse: Where Macaron Stands?

Author: Boxu Li at Macaron

Introduction: The Battle for Consumer Attention in the Age of AI

Artificial intelligence has moved from laboratories into the heart of consumer life. Platforms like ChatGPT have shown that large language models can answer questions, draft essays and plan trips. Now, OpenAI, the company behind ChatGPT, is pushing even further into the consumer realm with a pair of ambitious products: a TikTok‑style app powered by Sora 2 and a personalized daily briefing called ChatGPT Pulse. These experiments reveal how AI companies hope to capture more of our attention and how they seek to build stickier ecosystems around generative models. For Macaron, a nimble AI assistant focused on everyday living, understanding these moves is critical to charting its own path.

This article provides an erudite, evidence‑backed analysis of OpenAI’s new consumer products. We examine the technical capabilities of Sora’s video generation and Pulse’s proactive research, explore the broader consumer ecosystem OpenAI is constructing, and connect these findings to the emerging strategy of Macaron. By situating each product within current market and financial realities, we offer insights into why Macaron’s daily‑life‑centric approach may ultimately resonate more deeply with users seeking trustworthy AI companions.

The Rise of AI‑Generated Video: Inside Sora 2’s TikTok‑Style App

From Model to Mobile Feed

OpenAI’s Sora 2 is an upgraded video‑generation model designed to create realistic clips based on text prompts. Rather than deploying the technology solely through APIs or professional tools, OpenAI plans to release it in an app that mimics TikTok’s vertical feed. According to an Engadget report that draws on internal documents, the app will feature a vertical video stream with swipe‑to‑scroll navigation[1]. Users cannot upload footage from their camera roll; every video in the feed is produced by the AI itself[1].

This move signals a shift from AI as an on‑demand tool to AI as content creator. By controlling the supply of videos, OpenAI hopes to establish a wholly new medium populated by synthetic creativity. Each clip is limited to ten seconds, a constraint likely imposed to manage compute costs and to maintain snack‑size entertainment[1]. Users can like, comment on and remix these clips, and the app includes a “For You” page driven by recommendation algorithms[1]. In essence, it replicates the social dynamics of TikTok while swapping user‑generated content for machine‑generated visuals.

AI making AI generated content for human to consume. This visual is meant to tease the AI‑only nature of the new app, where every clip is born from a prompt rather than a camera.

Identity Verification and Deepfake Safeguards

One of the most novel aspects of the Sora app is its identity‑verification system. Before their likenesses can appear in generated videos, users must verify their identity through a face scan. Once verified, Sora can generate clips featuring them, and other users can remix those clips. Whenever someone’s likeness is used, they receive a notification[2]. This approach attempts to address one of the central dilemmas of generative video: how to prevent unauthorized deepfakes while allowing personalized content.

OpenAI has also baked in copyright protections. According to Reuters, rights holders will need to opt out if they do not want their content used for training or generation[3]. The app refuses to generate videos when prompts trigger copyright or safety filters[4]. Furthermore, it will not generate images or videos of public figures without permission[3]. These safeguards may limit some creative possibilities but are essential to avoid the legal and ethical quagmires that have plagued deepfake technology.

Employee Feedback and Productivity Question

Although the Sora app has not yet launched publicly, internal testing has been enthusiastic. Wired reports that OpenAI employees use the tool so frequently that some managers joke it could hurt productivity[5]. Such intense use hints at the addictive potential of AI‑generated entertainment, but it also raises questions. Will an endless stream of machine‑generated videos enhance users’ lives or merely divert attention? Unlike TikTok, Sora’s app lacks human creativity and unpredictable content, which could limit its staying power if videos start to feel repetitive.

Competitive Landscape and Opportunity Windows

OpenAI’s decision to mimic TikTok is not made in isolation. Meta’s Vibes feed and Google’s incorporation of its Veo 3 model into YouTube show that major tech players are racing to integrate AI‑generated video into social experiences. OpenAI appears to be moving quickly while TikTok faces regulatory pressures in the United States[6]. If lawmakers force ByteDance to divest its U.S. operations, an AI‑only feed may capture some displaced users. However, the absence of human creators means the app must rely entirely on the novelty and quality of AI output to keep audiences engaged.

Copyright, Moderation and Ethics: The Difficulties Behind AI Video Platforms

OpenAI’s Sora strategy underscores how complicated it is to scale AI video generation responsibly. The company’s opt‑out policy for rights holders may comply with existing laws, but it shifts the burden onto creators to keep their works out of AI models. At the same time, the ten‑second clip limit suggests that longer videos may be computationally expensive[1]. To support a full‑fledged social platform, OpenAI will need massive GPU resources, contributing to the company’s already extraordinary burn rate.

Moderation is another hurdle. TikTok spends billions on content moderation and still faces criticism for failing to filter harmful videos. In an AI‑only environment, moderation must address not only user comments but also the outputs of the model itself. OpenAI will need advanced filters to prevent the generation of violence, misinformation and abusive content. The success of the Sora app will therefore hinge not just on the model’s creativity but on the robustness of its guardrails.

ChatGPT Pulse and Macaron

Beyond Chat: How Pulse Works

OpenAI’s second major consumer experiment is ChatGPT Pulse, a feature that proactively delivers a personalized set of updates to users. Pulse analyzes a user’s chat history, memory and optional connections to services like Gmail and Google Calendar. Each night the model performs asynchronous research, gathering news, reminders and summaries on topics the user cares about[7]. The next morning it presents the findings as a series of visual cards that can be expanded into full conversations[8].

Pulse is deliberately finite. Unlike social feeds that encourage endless scrolling, Pulse refreshes once a day and the cards disappear unless saved[9]. Users can provide thumbs‑up or thumbs‑down feedback on cards and use a “curate” button to shape future updates[7]. This design aims to provide actionable information without creating an addictive loop.

A Shift Toward Agentic AI

OpenAI’s executives have positioned Pulse as the first step toward agentic AI. Traditionally, ChatGPT has been reactive: users ask questions and receive answers. Pulse changes this dynamic by having the model anticipate what the user might need. As VentureBeat explains, Pulse is a continuation of OpenAI’s earlier “Tasks” feature, which allowed users to schedule specific actions. Pulse automates that concept by having the assistant independently gather information[10]. The company envisions future agents that can research, plan and take actions without waiting for explicit instructions[11].

From a user‑experience perspective, Pulse resembles Apple’s Siri suggestions, Google Now and even news‑aggregation apps like Flipboard. Tom’s Guide notes that Pulse provides a morning briefing tailored to your goals and habits, offering restaurant recommendations, meeting agendas and relevant news[12]. It’s designed to slot into existing routines, encouraging you to check ChatGPT first thing in the morning.

Access and Equity: Who Gets Pulse?

Currently, Pulse is available only to ChatGPT Pro subscribers, who pay roughly US$200 per month[13]. OpenAI intends to gather feedback before expanding to Plus tier users and eventually to free users. This pricing creates an exclusivity barrier. While enterprise customers and power users may appreciate the convenience, the high cost limits the feature’s reach. Moreover, because Pulse draws on personal data, it raises privacy concerns. Some early observers worry about the potential for echo chambers and AI‑driven filter bubbles[14]. For Macaron, which targets a broader consumer base, such barriers highlight an opportunity to deliver proactive assistance without gating features behind expensive tiers.

The Consumer Ecosystem & Financial Realities

A Sprawling, Expensive Platform

Between Sora, Pulse, voice assistants, custom GPTs and the GPT Store, OpenAI is assembling a comprehensive consumer platform. The GPT Store, launched in January 2024, offers a marketplace of mini‑apps built by partners and the community. OpenAI announced that users have created more than three million custom GPTs and that the store includes categories ranging from writing and research to programming and lifestyle[15]. Yet the store is accessed through the ChatGPT interface and lacks the dedicated discovery tools of mature app stores. For many users, ChatGPT remains a chat interface first and an app ecosystem second.

OpenAI’s diversification is expensive. Industry analyses suggest the company lost about US$5 billion on US$3.7 billion in revenue in 2024 and expects to burn US$8 billion in 2025[16]. Despite hitting US$12 billion in annual recurring revenue in 2025, OpenAI does not anticipate being cash‑flow positive until 2029, aiming to generate about US$2 billion in cash that year[16]. Business Insider further reports that internal projections forecast cumulative losses of US$44 billion between 2023 and 2028[17]. Much of this spending goes toward compute and data‑center investments necessary to run models like GPT‑4 and Sora.

The Push for User Engagement vs. User Trust

Both the Sora app and Pulse reveal an implicit goal: capturing more user time. In the hyper‑competitive attention economy, platforms that keep users engaged can monetize through subscriptions, upsells or advertising. Yet there is tension between engagement and trust. For example, Pulse’s reliance on user data could create new privacy worries. Meanwhile, Sora’s AI‑only feed might be perceived as synthetic entertainment rather than genuine expression. Macaron’s challenge and opportunity is to strike a balance—delivering useful assistance without overstepping into manipulation or surveillance.

Why Macaron’s Daily‑Life Focus Could Win

Deep Domain Focus vs. Generalist Ecosystems

Macaron is not trying to be an “everything platform.” Instead, it aims to excel at daily‑life assistance—meal planning, grocery management, family scheduling and local recommendations. This focus allows for specialized models that understand dietary restrictions, household budgets and personal preferences. In contrast, OpenAI’s generalist models are optimized for broad tasks. While they can generate recipes or suggest schedules, they may lack the fine‑tuning required to provide contextually relevant, culturally appropriate advice.

Simplicity and Usability

Where OpenAI’s suite of products forces users to navigate chat, voice, video and mini‑apps, Macaron can deliver a single, unified interface built around daily routines. The cognitive load of switching between ChatGPT, Sora, the GPT Store and various connectors can frustrate users. Macaron can avoid that complexity by keeping core features front and center and by ensuring that every interaction adds value to daily life.

Agility and Customer Feedback Loops

As a smaller company, Macaron can iterate quickly based on user feedback. OpenAI must consider regulatory constraints, brand risk and the demands of its enterprise partners. In contrast, Macaron can pivot swiftly when a feature doesn’t resonate or when users express new needs. This agility is particularly valuable as generative AI capabilities evolve rapidly.

Privacy by Design

Macaron can differentiate itself by minimizing data collection. Pulse requires access to users’ email and calendar data to deliver relevant cards[7]. Sora’s feed gathers behavioral signals through likes, comments and watch time. Macaron can adopt a privacy‑first stance, asking for only the information necessary to deliver a service and providing clear options to delete data. Building trust at this stage of the AI boom may prove more valuable than being first to market with every new feature.

Early‑Stage Opportunity

The consumer AI market remains unsettled. While OpenAI has a head start, it is still experimenting to find products that stick. Sora’s 10‑second video limit and Pulse’s Pro‑tier exclusivity reveal that these products are not yet mass‑market offerings. Macaron can seize this moment to deliver immediate, tangible value to a broad audience without waiting for multi‑year roadmaps. By focusing on essential tasks and delighting users daily, Macaron can secure a loyal user base before AI feeds become saturated.

Conclusion:

OpenAI’s TikTok‑like Sora app and ChatGPT Pulse showcase the company’s ambition to dominate the consumer AI space. Sora presents a bold experiment in AI‑only entertainment, combining cutting‑edge video generation with social‑media mechanics. Pulse pushes ChatGPT toward proactive assistance, delivering daily summaries and nudging users toward a more agentic relationship with AI[11]. However, these innovations also surface difficult questions about content moderation, copyright, privacy and the role of AI in our attention economy.

Macaron’s strategy diverges by honing in on daily life. Rather than building an expansive ecosystem, it seeks to become the trustworthy companion that simplifies cooking, shopping and scheduling. This focus yields advantages in domain expertise, interface simplicity, agility and privacy. As the AI landscape matures, users may gravitate toward assistants that feel helpful and unobtrusive rather than platforms that compete for every spare second.

For a startup like Macaron, the path to a billion‑dollar valuation will not come from replicating the scale of OpenAI. It will stem from delivering consistent, trusted value to everyday people. By learning from OpenAI’s experiments and avoiding their pitfalls, Macaron can carve out a niche that remains relevant as AI permeates every facet of our digital lives.

[1] [2] [4] [6] OpenAI will reportedly release a TikTok-like social app alongside Sora 2

https://www.engadget.com/ai/openai-will-reportedly-release-a-tiktok-like-social-app-alongside-sora-2-205842527.html

[3] OpenAI's new Sora video generator to require copyright holders to opt out, WSJ reports | Reuters

https://www.reuters.com/technology/openais-new-sora-video-generator-require-copyright-holders-opt-out-wsj-reports-2025-09-29/

[5] OpenAI Is Preparing to Launch a Social App for AI-Generated Videos | WIRED

https://www.wired.com/story/openai-launches-sora-2-tiktok-like-app/

[7] [8] [11] [13] ChatGPT Pulse delivers morning updates based on your chat history - Ars Technica

https://arstechnica.com/ai/2025/09/chatgpt-pulse-delivers-morning-updates-based-on-your-chat-history/

[9] [12] ChatGPT Pulse is here to challenge Google News and Flipboard — now the AI starts the chat and curates your feed | Tom's Guide

https://www.tomsguide.com/ai/chatgpt-pulse-is-here-now-ai-starts-the-chat-and-curates-your-feed

[10] ChatGPT Pulse delivers daily personalized research, moving AI from reactive to proactive | VentureBeat

https://venturebeat.com/ai/chatgpt-pulse-delivers-daily-personalized-research-moving-ai-from-reactive