Proposed #deepfake law (or “falsul se(r)ver”, as the bill’s initiators translated this term) went from a whimsical form, in the initial version submitted to the Senate, to a relatively acceptable text after its adoption in the Senate, for the adoption of the “Chinese” version in the committees of the Chamber of Deputies – China is the only country in the world with specific tough #deepfake obligations and detailed obligations to flag this type of content.

The text adopted by the committees of the Chamber of Deputies in December 2023 and the vote on it at the plenary session on Monday, February 12, 2024, put #deepfake AI in the same definition as generated content “by using reality virtual” (?!? – this is exactly the text in the law!). which would create “the visibility of what a person said or did what in reality nothing was said and nothing was done”. But, gentlemen, NOTHING is real in virtual reality! So about whosereality“do you speak

In addition, from the wording of Art. 3 of the proposed law means that any #deepfake material can be legally used anywhere, even if it is “maliciously“and will”creates information chaos“, if marked in accordance with the law:

Finally, the text regulates distribution deep fake in the mass media – a term that is not legally defined anywhere in the legislation, but instead completely ignores the distribution deep fakes on other Internet channels (which do not fall within the category of online platforms as defined by the provisions of the Digital Services Act DSA) – for example a website that is not a publication mediator or a blog.

Instead, the text of the law goes into unnecessary detail by regulating the exact minimum percentage that the warning sign must take up, seemingly not realizing that 10% of the image can be manipulated in any way so that a portion is not visible de facto (find here an example in this sense developed by us).

As for the offenses added by the new law, they suffer from the same absurdity as those listed above. In essence, the new law aims to regulate a separate situation of the crime already provided for in Art. 362 CC – Changing the integrity of computer dataonly the criminal sanction would be milder than the current one. Perhaps this is exactly what I wanted!

The whole approach now seems to focus on recent problematic situations – deep fakes with politicians promoting scams through advertisements spread by online platforms, but seeming to completely ignore the much more practical and problematic situations that primarily affect ordinary users (eg revenge porn videos).

Furthermore, the entire bill seems to ignore – although it was used in the argument for the amendment – that deep fake– problematic ones will be regulated in the EU by the Regulation on Artificial Intelligence (AIA), which is in the last hundred meters before adoption. The regulation proposes a technology-neutral text with exceptions to freedom of expression and no desire for over-regulation – in stark contrast to the Romanian lawmaker’s text.

Solutions to Fake Content, Misleading and Online Fraud – Author deep fakes or anything else – they can only be provided by competent institutions with clear mandates that work transparently and effectively in particular with “online intermediaries” (as defined in the Digital Services Act) who must be held accountable if they do not act quickly in obvious cases

But for that you have to know what you are talking about. Because the main problem – and which must be solved first of all – is to stop the distribution of illegal content. And this was not the case in recent weeks, as we all saw.

But it’s much easier to promulgate a new law (be it stupid) than to have competent institutions and functional procedures, right? _Read the rest of the article and comment on Contributors.ro