The next evolution, the Generative Ads Recommendation Model (GEM), aims to fully automate ad creation. Marketers will simply provide an image and a budget, and the AI will generate the entire ad library. This shifts the marketer's primary value from ad creation to optimizing the post-click customer journey and offer.

Instead of selling AI directly to consumers, Meta provides AI tools to its 15 million business advertisers. This makes ads smarter and more effective, increasing ad revenue. This profitable ad machine then funds Meta's massive, long-term AI ambitions, creating a powerful flywheel.

Ben Thompson argues AI apps should adopt a Meta-style advertising model based on deep user understanding, rather than Google-style contextual ads tied to prompts. This avoids conflicts of interest and surfaces products users didn't know they needed, creating more value for both users and advertisers.

An opt-in feature allows Facebook's AI to access your camera roll to suggest and create content like collages or videos. While this can rapidly generate posts from business events, it requires marketers to weigh the significant privacy implication of giving Meta deeper access to their raw photo and video data.

Meta's Muse Image model is being deeply integrated into Instagram and WhatsApp, allowing users to tag friends and insert their public photos into AI generations. This leverages the network effect to accelerate adoption, accepting the risk of 'one-click deepfake' controversy as a cost of viral growth.

Unlike competitors who would struggle to introduce ads into AI chat, Meta's user base is already accustomed to ads in their feeds. This gives Meta a unique advantage to monetize a proactive consumer AI agent that can surface sponsored suggestions for shopping or travel without creating user friction.

Major AI chatbots are designed with a default setting that opts users *into* having their conversations—including sensitive data—used for model training. This "opt-out" privacy model places the burden on the user to navigate settings and protect their own data, a critical fact many are unaware of.

An interaction with Meta's new AI demonstrates the fine line between helpful personalization and invasive creepiness. The AI suggested "Malibu appropriate surf puns" based on the user's private data (likely from Instagram), then awkwardly denied it. This highlights the PR and user trust challenges of leveraging personal data, even for seemingly innocuous features.

A new Marketing API feature allows Meta's AI to allocate up to 5% of ad spend to placements explicitly excluded by an advertiser. This signifies a major shift towards autonomous campaigns, reflecting Meta's confidence that its system can identify performance opportunities even in channels that human advertisers have ruled out.