Meta has just unveiled a new suite of artificial intelligence tools for Facebook and Instagram, marking another step in the race for generative assistants. The features, announced in the past few hours, focus on two main areas: advanced photo editing and a chatbot capable of answering complex questions. While these innovations promise to make the user experience smoother and more creative, they also raise significant questions about data management and the growing reliance on increasingly pervasive algorithms.
The most eye-catching element is undoubtedly the new photo editing tool. Integrated directly into the Facebook app, it allows users to edit images using text commands: you can ask it to remove a background, change an object's colors, or even generate new elements within the photo. This system is based on Meta's proprietary generative model, which the company says was trained on a vast dataset of public images. The ease of use is impressive, but it is legitimate to wonder how Meta will handle photos uploaded by users: will they be used to further train the models? Meta has stated it will follow existing privacy policies, but many observers remain skeptical, especially after recent controversies related to facial recognition in Ray-Ban Smart Glasses.
Sponsored Protocol
At the same time, the new conversational assistant, internally nicknamed "Meta Brain," is designed to answer questions about restaurants, travel, shopping advice, and much more, drawing not only from public Facebook and Instagram data but also from web-indexed information. This move closely resembles what ChatGPT and Google Gemini already do, but with an advantage for Meta: direct access to the profiles and preferences of its billions of users. Instead of just giving generic answers, the assistant can personalize suggestions based on user history, likes, and visited places. A concrete example is if you ask "where can I eat a good pizza tonight?" the system might suggest the very restaurant your friend reviewed positively last week.
Sponsored Protocol
Comparison with Existing Alternatives
Despite the promises, many analysts point out that these features do not represent a true innovation. AI photo editing is already available on tools like Adobe Photoshop and Google Pixel smartphones. Conversational chatbots, moreover, are now a commodity. What makes Facebook's new tools unique is Meta's closed ecosystem, which combines the power of AI with a vast social database. However, this also raises risks related to privacy and manipulation. A recent report highlighted that 85% of IT teams claim to control AI agents, but only 42% know the actual owners of the data being used, a phenomenon known as Shadow AI that could extend to consumer tools as well.
Sponsored Protocol
Moreover, Meta seems to be replicating what was already done with Facebook's previous AI mode, launched a few months ago for search and answers across Meta's public data. That feature, as detailed in our dedicated article, sparked a heated debate on the use of user data. The new photo editing and Q&A tools follow the same line, further expanding Facebook's AI analysis capabilities. For a broader perspective on how AI is concentrating wealth and power, read our analysis on growing inequality in AI.
From a technical standpoint, the new assistant seems to adopt a hybrid approach: combining information retrieval (retrieval augmented generation) with text generation to deliver relevant answers. However, the real difference will be made by user adoption. While the photo editing tools might attract content creators, the chatbot risks being perceived as yet another intrusion. Wikipedia reminds us that generative artificial intelligence raises profound ethical questions, and integration into social media amplifies them enormously.
Sponsored Protocol
In conclusion, Meta is playing a crucial game: it wants to prove it can compete with OpenAI and Google on the consumer AI front, leveraging its user base. But the road is uphill, and privacy concerns might dampen enthusiasm. The coming months will tell us whether these new features will be a success or yet another attempt to monetize personal data.
