Facebook's AI Backfires

Meta is trying to bring AI into its products when it is behind its competitors. Photo: Profit .

Although it has appeared in the Vietnamese version since the end of 2024, recently many users have noticed the frequent appearance of Meta's AI technology. From Facebook posts to Instagram and Messenger, purple circles continuously ask users if they need help or information interpretation.

In fact, this is a strategic move in the fierce AI race between technology giants. While OpenAI is holding the upper hand with ChatGPT, Google is strongly developing Gemini, Meta is actively promoting AI on the ecosystem to take advantage of the huge platform with billions of users to catch up with competitors.

However, this move by Meta also raises questions about information security. Some expressed annoyance at the technology being so deeply embedded in people's daily lives.

What is Meta AI?

This is a virtual assistant model developed by Meta, Facebook's parent company, based on the large language model (LLM) Llama3, and the latest is Llama 4. Meta considers Llama 4 an important step forward in the field of artificial intelligence, and is about to launch Llama 4 Behemoth - the world's smartest LLM.

Meta AI is now integrated into Facebook, Instagram, Whatsapp, and Messenger. You can use AI by creating a new conversation or mentioning @Meta AI in an ongoing conversation. The tool has been available in Vietnamese since October 2024.

In addition to the features of answering questions, suggesting content, and “Imagine” to create photos, Meta AI can help users save time or support creative work such as creating posts, editing photos, and searching for information directly. You can visit Meta AI's fanpage to refer to more applications of the tool.

Applications of Meta AI on the ecosystem. Photo: Facebook, Instagram.

Users can search for Reels related to the topic of conversation right in the message box. If ChatGPT is taking social networks by storm with the ability to create characters from toy boxes, users can also create and personalize AI characters according to their preferences on Meta AI.

From early 2025, Meta AI will be customized for Vietnamese users, with the ability to understand local culture and display more frequently. On Meta's ecosystem, just scroll through a few posts or enter the chat section and the tool's purple circle will appear.

By proactively integrating AI heavily, Meta not only wants you to use it more, but also creates an ecosystem where AI supports everything from search, shopping to social interactions, thereby increasing the value of these platforms.

User Response

While bringing a lot of entertainment value and convenience to users in applications, Meta AI is receiving many reviews about information security issues as well as the frequency is too dense, causing many people to feel overwhelmed.

Imagine you’re having a conversation with a friend, say about a movie you saw yesterday or something delicate that you need advice on. While it would be handy to have an AI explain the plot of the movie or articulate the point you want to make, it would detract from the value of the conversation.

The Guardian 's Polly Hudson was struck by the blue-purple circles, and suggested the reason for the online outrage was that they couldn't turn them off.

“We should be allowed to turn this feature off, rather than being forced to accept it. It is a menacing presence in the corner of the screen, always present, watching, waiting,” she wrote.

Chatting with each other using AI feels like two robots talking. Photo: The Guardian.

Previously, many newspapers such as Wired and AP News responded to Meta's use of user data to train their AI. Previously, Meta was also banned from releasing AI in the European market due to privacy violations. The parent company of a series of applications such as Facebook, Instagram, Messenger also admitted this.

Meta has now committed to only using publicly available and licensed information, and not using data from private messages or from users under the age of 18. The company has also open-sourced Llama 4 to increase transparency by allowing the community to participate in the process of customizing and developing the model.

However, the Open Source Initiative declared Llama 4 no longer open source in 2023. Meta has yet to fully disclose how it collects and uses user data to train its AI.

In the future, AI applications will become more and more popular. Users can take advantage of its benefits, but also need to verify information and protect personal data appropriately.

Source: https://znews.vn/ai-cua-cha-de-facebook-gay-phan-ung-nguoc-post1543996.html