Recently, the images created by the tool Midjourney where you can see the Pope Francisco wearing an extravagant Balenciaga brand coat or the photos of a false arrest of Donald Trump raised questions regarding the difficulties of readers or users to verify the veracity of the content they observe on networks.
the circulation of highly realistic fake images created by Artificial Intelligence It supposes a greater danger than what has even been evidenced regarding the possibility of increasing the waves of disinformation of which those who use the platforms are victims today.
That is why the Instagram social network, owned by Meta, announced that it will implement a series of resources to identify these cases and prevent users from being victims of fake news or AI-altered images.
To do this, the company reported that those publications that contain this feature will see a filter on the image with the following message: “Photo altered. Checkers independent data reviewers have already reviewed the same altered photo in another publication.”
In this sense, from the company they indicated that currently “if the checkers When third-party feeders spot false information, altered content, or content that lacks context on Instagram, we make it harder to find.”
But in addition, the company has a system of tags used to identify cases of false information. When faced with said label, the user can decide whether or not to be exposed to the content, and choose or not to share it.
Among the methods used is the reduction of the effectiveness of hashtags, in addition to preventing them from appearing in the section To explore e stories.
More Meta Changes
Meta, The company, which owns Facebook, published an artificial intelligence model on Wednesday capable of identifying individual objects within an image, alongside an image annotation dataset that it says is the largest of its kind.
The company’s research division claims in a blog post that its model Segment Anything Modelor SAM, can identify objects in images and videos even in cases where it hasn’t found them during its training.
With SAM, objects can be selected by clicking on them or by typing text prompts. In one demo, typing the word “cat” made the tool draw boxes around each of the cats in a photo.
Big tech companies have been hyping up their advances in artificial intelligence ever since the chatbot ChatGPT the OpenAI, supported by Microsoft caused a sensation in the autumn, triggering a flurry of investment and a race for industry domination.
Meta announced several features that implement the type of generative AI popularized by ChatGPT, which creates entirely new content rather than just identifying or categorizing data like other AIs, though it hasn’t released any products yet.
Some examples are a tool that creates surreal videos from text messages and another that generates children’s book illustrations from prose.
The CEO Mark Zuckerberg He said incorporating these generative AI “creative aids” into Meta apps is a priority this year.
Meta already uses SAM-like technology internally for activities like tagging photos, moderating prohibited content, and determining which posts to recommend to Facebook and Instagram users. The company claims that the launch of the SAM will broaden access to that type of technology.
The SAM model and dataset will be available for download under a non-commercial license. Users who upload their own images to an attached prototype must also agree to use them for research purposes only.
