Meta has begun deploying an AI assistant capable of performing tasks without constant human prompts, while simultaneously preparing to test a redesigned Facebook app that defaults to a full-screen, TikTok-like video feed. The twin moves, announced Friday, come after the company disclosed a combined loss of roughly 20 million users across its platforms in the first quarter of 2026, intensifying pressure to recapture engagement.

The updated Meta AI, powered by the new Muse Spark 1.1 model, can understand user context and execute recurring tasks—such as compiling daily summaries from calendar events or generating weekly meal plans—autonomously. Initial availability is limited to select markets via the dedicated Meta AI app and website, with plans to expand to WhatsApp and other platforms. Users retain control over interactions, and anonymous chat modes remain available for private conversations.

On the social network front, Facebook will begin testing a “reimagined experience” this fall in video-heavy countries, which opens directly into a continuous stream of vertical clips, mimicking the core mechanic of ByteDance’s TikTok. Tom Allison, the Meta executive overseeing Facebook, said the test group will be able to opt out and return to the traditional feed. A wider rollout, potentially including the United States, could follow in 2027. The shift acknowledges that video now drives conversation, community, and commerce, and responds to a user base that increasingly consumes rather than posts content.

To bolster its commercial ecosystem, Meta also launched Seller, a standalone app for Facebook Marketplace merchants, offering AI-powered listing creation, unified buyer communications, inventory management, and performance insights. Alongside it, the company introduced Facebook Verified, a free profile badge confirmed through a selfie check, aiming to address authenticity concerns.

The push into autonomous AI and algorithmic video feeds unfolds against a broader industry debate over risk. Elon Musk, promoting his own xAI venture, put the probability that artificial intelligence turns destructive at 10–20 per cent, arguing the potential for an “era of abundance” justifies the gamble. Meta’s chief executive Mark Zuckerberg countered with an optimistic vision of AI as an empowering tool, underscoring the company’s commitment to distributing benefits widely. Meta’s second-quarter earnings, due after markets close on 29 July, will offer the next concrete measure of whether these product bets can reverse user attrition.