A Manifesto That Sparks Trust Issues

Meta CEO Mark Zuckerberg released a roughly 6,500-word article this week titled The Future is for Everyone, claiming that everyone will soon have a “remarkably powerful” personal AI agent capable of understanding users, their goals, and everything they care about. But during a discussion on TechCrunch’s Equity podcast, several editors suggested that this vision hasn’t won everyone over.

“AI agents,” in simple terms, are more than just chatbots that answer questions—they’re software assistants that can plan, organize, write, and even execute tasks on behalf of users. Zuckerberg’s vision is one where AI becomes closer to personal devices and daily life, rather than remaining confined to cloud-based models or enterprise tools. On the surface, this sounds like a technology pledge aimed at the masses. But that’s precisely where the controversy begins.

Meta’s Bid for the “Personal Empowerment” Narrative

During the podcast, Rebecca Bellan suggested that Meta might be competing for position in the AI landscape through a different angle. She pointed out that Meta isn’t widely seen as a leader in the closed frontier-model space, nor does it necessarily hold an absolute advantage in the open-model race. So Zuckerberg chose to focus on “personal empowerment,” emphasizing that people could run their own personal AIs on their own devices.

Zuckerberg’s envisioned new Meta AI model, Glimmer, would be used to manage schedules, draft messages, and organize files—and it would be “always on,” running anytime, anywhere, even offline. Meanwhile, Muse Spark would allow Meta to retain some control over its most powerful models while creating a revenue outlet for individuals and businesses that need additional compute resources for larger projects.

Key takeaways include:

• 6,500-word vision piece: The central slogan is “the future belongs to everyone”;
• Personal AI scenarios: scheduling, messaging, file management, and more;
• On-device ambitions: previously linked to glasses and wearable devices;
• Accessibility questions: podcast guests noted that downloading and trying it on a MacBook isn’t possible—it requires specific hardware.

This means Meta is selling a future that’s “accessible to everyone,” but at this stage, the entry points, hardware forms, and whether ordinary users can get started with low friction remain unclear.

It’s Not Just the AI People Are Skeptical About—It’s the Speaker

The other layer of skepticism comes from Meta and Zuckerberg’s historical baggage. Bellan reflected on how, during the social media era, Zuckerberg also championed the idea of giving everyone a way to connect with friends and build social networks—yet what people often ended up with was anger-inducing content, advertising, and insufficiently genuine connections.

TechCrunch AI editor Russell Brandom went so far as to call the manifesto “exactly the kind of thing that makes people dislike AI.” Kirsten Korosec similarly argued that the piece misses the right note for many: it paints AI as a great tool for humanity without offering a more grounded picture or adequately addressing the significant costs that could come with it.

The “costs” here aren’t limited to the risks of any single product—they reflect a broader public unease with large tech companies: How is data being used? How will the platform monetize? Will these technology promises become gateways for advertising and control? Meta’s Llama was praised by podcast guests as an excellent open-source tool, but in terms of consumer-grade experience, many users’ impression of Meta AI still centers on chatbots within social platforms rather than a personal assistant they’d trust with their day-to-day affairs.

An “Anti-Dario” Stance and the AI Industry’s Divide

The podcast also drew a contrast between Zuckerberg and Anthropic CEO Dario Amodei, who recently spent time addressing whether he’s an “AI doomer.” In Bellan’s view, Zuckerberg appears to be cultivating an “anti-Dario” posture: rather than advocating for slowing down, he’s arguing that we can’t afford to stop—and even that decelerating would harm the individuals who stand to benefit from AI empowerment.

This reflects a core split in the current AI industry: on one side, frontier labs publicly stress safety, boundaries, and caution; on the other, large platform companies want to turn AI into consumer products, device entry points, and commercial ecosystems as fast as possible. The so-called “frontier models” refer to the most capable large models in the industry, which typically entail enormous training costs and computational demands. Despite massive investment—and previous headlines about poaching AI scientists at high salaries—Meta still isn’t the first company most people think of when they imagine using a personal assistant.

Industry Take: Visions Need to Be Validated by Products and Trust

This debate underscores that AI companies can no longer win users over with grand declarations alone. People aren’t entirely rejecting personal AI: schedule management, writing assistance, and file organization all have genuine practical value. But when these features are packaged as a future that “benefits everyone,” users naturally ask: Who controls the model? Who owns the data? Who profits? And can ordinary people actually access it?

Meta’s next challenge isn’t just about model capability—it’s about accessibility and credibility. If “the future belongs to everyone” still depends on specific hardware, closed entry points, or opaque business logic, it will struggle to shake off the shadows of the platform-company narratives of the past. Personal AI assistants may well become the next major entry point, but whoever wins will be the one that delivers useful functionality, clear boundaries, and user trust—not the one that tells the most thrilling story about the future.