Zuckerberg wants to personalise your digital future — but will it be yours?

“Mark Zuckerberg just pitched a future where everyone has their own AI. It sounds empowering. But it might be the most important power grab of the decade.”

When Meta’s CEO announced a plan to build “personal superintelligence for everyone,” it landed like a thesis statement for the next computing era: ambient assistants embedded in apps, glasses, and devices that learn your preferences, anticipate needs, and act on your behalf. The promise is undeniably compelling — and with Meta’s distribution across Facebook, Instagram, WhatsApp, and its growing wearables footprint, the company has the reach to make it happen at unprecedented scale. (About Facebook)

But scale cuts both ways. The same infrastructure that can deliver convenience can also consolidate power — over your data, your attention, and ultimately your choices. If the “AI for everyone” future is built on closed stacks, captive identity, and data extraction by default, we won’t own our agents. They’ll own us.

This is the debate we need now: What makes an AI truly personal? And how do we ensure it remains yours?

Meta’s vision meets real-world frictions

Meta isn’t just talking about software. It’s building the hardware beachhead for ambient AI through Ray-Ban Meta smart glasses — an increasingly popular form factor with multimodal assistants you can speak to and (soon) gesture with, while a camera sees what you see. Analysts report Meta’s outselling rivals and tightening ties with EssilorLuxottica to control the pipeline from design to retail. (WIRED)

That ubiquity raises new social questions. A recent wave of public pushback — led by privacy-minded Gen Z — highlights discomfort with being recorded in everyday spaces, especially as tutorials circulate on how to bypass recording indicators. Even if the tech complies with policy, social norms are signaling a desire for stronger boundaries and consent. (The Washington Post)

Meanwhile, regulatory pressure is shaping how “personal” these AIs can be — especially in Europe. Meta paused plans to train its models on EU and UK user data following interventions by privacy regulators, underscoring that “reach” is not the same as “permission.” (TechCrunch, About Facebook, The Hacker News) And the EU’s broader platform rules (DMA) and data-sharing laws (Data Act) are pushing toward portability and interoperability, so people — not platforms — determine where their data flows and which services can make use of it. (European Commission, Digital Strategy Europe, Tech Policy Press)

Bottom line: Personal AI at platform scale is not guaranteed to be person-first. The guardrails we choose — in product design and policy — will decide whether this future empowers users or entrenches gatekeepers.

A simple test for “personal AI”: Personal. Portable. Private.

If we want personal AI to serve people — not just platforms — it must pass three tests:

1. Personal (user-controlled identity and intent). Your agent should take actions as you, using credentials you control, with a transparent policy describing what it can do and why. That implies verifiable identity primitives for both humans and agents, explicit consent, and revocation that actually works. (Think: “Know Your Agent” alongside KYC — who’s acting, on whose authority, within what scope.)

2. Portable (works across apps, devices, and ecosystems). A truly personal AI should be wallet-portable and vendor-agnostic, not bound to a single app or device. EU regulation points this way; the DMA and Data Act emphasize data portability and interoperability so value (and rights) travel with the user. (European Commission, Digital Strategy Europe)

3. Private (minimize, prove, don’t expose). Sharing the minimum necessary proof — not raw data — should be the default. Privacy-preserving verification (e.g., selective disclosure, zero-knowledge proofs, on-device processing) can power useful behaviors without turning every interaction into more training data. Europe’s response to platform training plans is a cautionary tale here. (TechCrunch, About Facebook)

The architecture we need: identity as the control plane

To operationalize “personal, portable, private,” we need identity as the control plane for AI:

• Verifiable credentials in a wallet anchor who you are (and what your agent is allowed to do).

• Policy + permissions live with the user (not just the app): scopes, time limits, risk thresholds, human-in-the-loop requirements.

• Selective disclosure lets agents prove enough (age, residency, authorization) without oversharing.

• Interoperable attestations move across platforms so your AI can act in different contexts without re-KYCing your life.

• Revocation & audit give you a trail — and a kill switch.

This is where Validity Worldwide’s VALid™ fits — not as “anti-Meta,” but as a neutral trust layer that keeps identity and permissions with the person, even as assistants roam across apps and devices. In practice, that means:

• Wallet-anchored identity for humans and agents (KYA), so actions are attributable and accountable across ecosystems.

• Portable credentials that work across walled gardens — aligned with the direction of data portability and interoperability norms. (European Commission, Digital Strategy Europe)

• Privacy by design through selective disclosure, minimizing what’s shared while maximizing what can be verified.

If the next interface is eyewear and ambient devices — a race Meta currently leads — user-controlled identity becomes the line between augmentation and surveillance. (WIRED, The Washington Post)

Power, platforms, and permission

“Zuckerberg is right on one thing. This is the decisive decade. But we shouldn’t confuse resources and reach (which he has) with permission (which he doesn’t).”

Permission will be earned through architecture and conduct, not merely announced in a keynote:

• Build opt-in, not opt-out systems — especially for training data. Regulators have made clear that “publicly available” isn’t carte blanche. (TechCrunch)

• Make portability real: let users take their agents — identity, history, preferences, boundaries — to another ecosystem without starting over. (European Commission, Digital Strategy Europe)

• Treat identity as a user asset, not a platform entitlement.

“Meta may deliver ‘superintelligence’ to billions. But unless it’s truly personal, truly portable, and truly private, then we’ll just be users again.”

The opportunity: a personal AI that’s actually yours

We don’t have to choose between innovation and rights. The same techniques that secure money and critical infrastructure can secure our digital selves:

• Verifiable, wallet-based identity for people and agents

• Interoperable credentials across apps, devices, and jurisdictions

• Privacy-preserving proofs instead of raw data dumps

• Transparent permissions with real-time revocation

This is how personal AI stays personal — whether it’s living in your favorite app, your browser, or your glasses.

Meta’s vision matters. So does the public’s appetite for boundaries and the law’s insistence on permission. The decisive decade will belong to builders who deliver intelligence and agency — and who treat identity as the user’s, not the platform’s. (The Washington Post, European Commission, About Facebook)

Sources & further reading:

• Meta: “Personal superintelligence for everyone” announcement (July 30, 2025). (About Facebook)

• Meta pauses training on EU/UK user data amid regulatory pressure (June 2024). (TechCrunch, About Facebook, The Hacker News)

• EU Digital Markets Act (DMA) overview; EU Data Act explainer on interoperability and portability. (European Commission, Digital Strategy Europe)

Smart-glasses adoption and privacy backlash reporting. (WIRED, The Washington Post)