The social platform X (formerly Twitter) officially installed a new feature, Photo Editor, on December 24, allowing users to edit the images attached to the post directly through the Generated AI, and non-authors can use it to cause misperceptions, and the content of the service provisions, which were changed on December 16, has received renewed attention.
According to the bulletin, the new version of the service provision is expected to enter into force on 15 January 2026 and will remain in force until then.
The amendments cover a number of additions, in particular in Article 3 “Serious content”, which clarifys the attribution of responsibility for what is generated by the generation of AI and is considered to be closely related to new functions in the near future.
Generating AI Output Content Legal Accountability to Users
The current provisions provide that the users are responsible for their own use of the services and their published content and are subject to all applicable legislation. However, in the new article, the scope of responsibility is further specified by stating expressly that:
According to the official Chinese language clause, the user must ensure, on its own initiative, that the content it provides, establishes or generates is in conformity with all applicable laws, rules, policies and regulations, and that this obligation applies equally to services provided by third parties or X-related enterprises.
This is also read to include the “Grok” model of AI developed by xAI and integrated into X platform, and as long as the user is able to produce the content of the generated AI function, if it involves violations or abuses, the user will be solely responsible.
The Korean painter was angry with AI and decided to add all the images posted on Twitter to Mask’s face before the platform returned to its former state, “Ellomaske, let’s enjoy!”
In addition, the provisions of the “authorization of content rights” previously controversial have been adjusted in this revision. X Re-emphasizes that users still retain ownership of the content they publish, but at the same time need to grant X global, non-exclusive, royalties.
It is worth noting that the new article adds that not only the content uploaded by the user, but even what is created or generated by the service, is included in the mandate.
Future users may also be used for the operation of services, for dissemination, and for machine learning and AI model training, by means of images, text or other content produced through the generation of AI functions.
While users can choose to opt out of part of the information for training purposes in the set, there are still structural difficulties in the control of the right to create, with regard to content that can easily be reproduced and reused in large numbers, such as game clippings, illustrations, etc., which are still difficult to resolve.
In fact, in the previously launched AI Image Generation Service “Grok Imagine”, the “Make Video with Grok” function has been controversial, with users producing videos directly, without their consent, with photos of real people, on the right to image and ethical issues.
The “photo-editor” feature on the line this time is considered to further magnify the risks of hypocrisy and abuse. Since the platform does not currently provide a mechanism for the original author to prohibit others from using the feature to edit his picture, many creators fear that the work will be easily adapted and circulated.
While X emphasizes that “pictogram editing” does not directly replace images in the original post, but rather allows users to download them themselves or to publish the results of the editing separately, in practice, the effect is no different from the effect of unconsensual processing images through external AI.
In the new version of the article, X clearly states that the responsibility for the content generated through the generation of AI and involving violations or breaches of the law rests with the users themselves, which is equivalent to “the relaxation of the AI tool but the devolution of risks to the users ” .
