The discovery that the ads approved by X contains the hate speech that is hostile to violent Muslims, before the German elections.
Social media giants agreed to Meta and X ads that users in Germany used to hate anti -Muslims and anti -sale in the period before the country’s federal elections, according to a new research from EcoA group of non -profit campaign.
The researchers in the group tested whether the advertising review systems for the platform will agree or reject the submission of advertisements that contain hateful and violent correspondence targeting minorities before the elections, as immigration occupied the lead in the prevailing political discourse-including advertisements that contain anti-Muslim solas. He calls for immigrants prison in detention camps or to be gas; And the images created from artificial intelligence from the mosques and temples that are burned.
Most of the test ads were approved within hours of submitting the review in mid -February. Federal elections in Germany are scheduled to be held on Sunday, February 23.
Scheduled hate ads
EKO said that X agreed to all the ten hate ads who only presented researchers a few days before the federal elections, while Meta agreed to half (five ads) to operate on Facebook (and perhaps Instagram) – although it rejected the other five.
The reason indicated that Meta presented to the five novels indicated that the platform believes that there may be risks to political or social sensitivity that may affect the vote.
However, the five advertisements agreed upon by a dead man included a violent hate speech similar to Islamic refugees to “virus”, “Firmin”, or “rodents”, and described Muslim immigrants as “elaborate”, and they claim them to sterilize them, burnt or gas. Meta also agreed to an advertisement that calls for temples to be burned to stop the “global Jewish mice agenda”.
Aside, Eco says that there is no one of the images created from artificial intelligence that I used to clarify the ads of hate speech that was classified as artificially created-however, half of the ten ads were approved by Meta, regardless of the company That has A policy that requires detection of the use of artificial intelligence images For advertisements on social issues, elections or politics.
Meanwhile, X agreed to all these five ads – and five others containing the violent hate speech targeting Muslims and Jews.
These additional approved ads included attacking the attacking correspondence “rodents” that claimed that the advertising version that it claimed was “immersed” the country to steal our democracy. .
The last announcement was combined with the images created from artificial intelligence and depicted by a group of dark men sitting around a table surrounded by tools of gold bars, with a star of David on the wall above them-with visual images also firmly tends to anti-Semitic bangs.
The AD X, which was approved on a direct attack on the Democratic Democratic Party, included the center left, which is currently leading the coalition government in Germany, with the allegation that the party wants to take 60 million Muslim refugees from the Middle East, before this tried a fierce response. It is also scheduled to announce a “left -wing” announcement of “open borders”, and calls for the extermination of “rapists”.
Elon Musk, the owner of X, used the social media platform as he had nearly 220 million followers to intervene personally in the German elections. in Tweet in DecemberThe German voters called for the far -right AFD to support Germany. Livestream has also hosted with AFD leader, Alice Widel, on X.
EKO researchers disabled all the test ads before any of them were arrested to ensure that no users of the platform were subjected to violent hate speech.
She says the tests highlight the flagrant defects with the AD platform approach to moderate content. In fact, in the case of X, it is not clear whether the platform is doing any moderation in ads, given that all ten ten hate ads were approved for the display quickly.
The results also indicate that ads platforms can get revenue as a result of the distribution of violent hate speech.
European Union Digital Services Law in the framework
EKO tests indicate that there is no platform that properly carries out the ban on hate speech, as both claim to apply it to the advertising content in their own policies. Moreover, in the case of Meta, EKO reached the same conclusion after a procedure A similar test In 2023, before the rules of governance online in the new European Union, it indicates that the system has no effect on how it works.
“The results we have found indicate that the baptism systems that depend on the AI-Meeta are still mainly suspended, although the DEA (DSA) has now completely.”
“Instead of enhancing the process of reviewing advertisements or hate speech policies, the definition seems to be declining through the painting,” they added, referring to The company’s last announcement about the retreat of moderation and fact -verification policies As a sign of the “active slope” they suggested putting in a direct collision cycle with DSA rules on regular risks.
EKO presented its latest results to the European Commission, which oversees the DSA’s main application on the husband of social media giants. She also said that she is involved in the results with the two companies, but she did not respond.
The European Union opens DSA investigations in definition and xWhich includes concerns about election security and illegal content, but the committee has not yet concluded. Nevertheless, Once again in April She said she suspects insufficient definition of political advertisements.
A preliminary decision on part of DSA investigations on X, which has been announced In JulyDoubts included that the platform fails to raise the level of the rules of advertising of the organization. However, the full investigation, which started in December 2023It is also related to the illegal risks of the content, and the European Union has yet to reach any results on the largest part of the investigation after more than a year.
DSA’s confirmed violations can attract penalties up to 6 % of its global annual rotation rate, while systematic non -compliance may lead to access to platforms that are temporarily banned.
However, at the present time, the European Union still takes its time to compensate for Meta and X investigations for that – awaiting final decisions – any DSA sanctions are still in the air.
Meanwhile, the matter is now hours before German voters went to the polls-indicating an increasing group of civil society research that the organization of leading governance on the Internet in the European Union has failed to protect the main democratic process of the European Union’s economy from a set of technology Fit by technology threats.
Earlier this week, Global Witness Absolute The results of the X and Tiktok algorithm tests “for you” in Germany, which indicate that the platforms are biased to enhance the AFD content for the content of other political parties. The civil society researchers as well X accused of banning access to data To prevent them from studying security risks in the elections in the period before the German poll-it is supposed to be enabled to reach DSA.
“The European Commission has taken important steps by opening DSA investigations in both Meta and X, and we now need to see that the committee is taking strong measures to address the concerns that have been raised as part of these investigations,” Eco’s spokesman told us.
“The results we have reached, along with increasing evidence from other civil society groups, show that large technology will not clean its platforms voluntarily. Despite their legal obligations under DSA. (We blocked the name of the official spokesperson for harassment.)
“The organizers must take strong measures-whether in the DSA application, but also, for example, to implement mitigation measures before the elections. This can include stopping the operation of the recommendation-based recommendation systems immediately before the elections, and implement other appropriate measures from the appropriate” glass “to prevent the algorithm amplification For the content of the border, such as hating content in the elections issued.
The group of campaigns also warns that the European Union is now facing pressure from the Trump administration to alleviate its approach to organizing large technology. “In the current political climate, there is a real danger that the committee does not completely implement these new laws as a privilege for the United States,” they suggest.
