DSA: EU Commission investigates recommendation algorithms of TikTok and YouTube

On the basis of the Digital Services Act (DSA), the EU Commission sent a request for information to YouTube, Snapchat and TikTok on Wednesday this week. It is asking the platform operators to provide more information about the design and functioning of their recommendation systems. This concerns, for example, what content is displayed to users in their timelines and news feeds and for what reasons. According to the DSA, platforms must assess the risks posed by their recommendation systems and mitigate them appropriately.

This includes risks to users' mental health and the spread of harmful content. These could arise, for example, from interactions based on the structure of the relevant algorithms, the Commission explains. The three operators have until November 15 to respond.

In the case of YouTube and Snapchat, the Brussels-based government institution is initially requesting general information about the parameters used by their algorithms to recommend content to users. In particular, it is looking at the potential impact on the voting process of those affected, civil society discourse and the psychological well-being of users. The Commission is concerned about algorithmically fueled addictive behavior, for example. The inquiry is also aimed at the protection of minors and the dissemination of illegal content, including drug-related themes and massive hate speech.

GDPR proceedings are underway

In the case of TikTok, the executive authority is interested in how the manipulation of the service by harmful actors is prevented. In addition, the company, which belongs to the Chinese group ByteDance, is to explain exactly how it mitigates risks in connection with elections, media pluralism and civil society discourse that could be amplified by certain recommendation systems.

At the same time, the Commission threatens "the next steps" that it intends to take based on the assessment of the responses. This could result in the formal initiation of proceedings based on the General Data Protection Regulation (GDPR). The government body also points out that it can impose fines for incorrect, incomplete or misleading information in the responses. If those contacted do not respond at all, fines may be imposed.

Algorithms promote "disinformation and polarization"

Since the DSA came into force, the Commission has focused heavily on compliance with the provisions of the Platform Act on recommendation systems. They are also at the center of a similar request for information sent to Amazon in July. Shortly before that, the watchdog made similar requests to the Chinese e-commerce giants Temu and Shein. Relevant algorithms are also an important part of the ongoing infringement proceedings against TikTok, AliExpress, Facebook and Instagram.

Green MEP Alexandra Geese, who helped negotiate the DSA, welcomed the new step. She warned that the business model associated with the recommendation systems has long posed a threat to democracy, "because it enables the spread of disinformation and polarization". The algorithms used determine which content spreads particularly well on the platforms. So far, they have mainly rewarded hate, incitement and highly polarizing posts. In concrete terms, this means that such extreme posts "spread much faster than other content, first on the platforms themselves and then in other media, including talk shows"

Since the revelations of whistleblower Frances Haugen, it has been known, according to Geese, that Facebook, for example, was aware of these effects. Despite this, the network, which belongs to Meta, has not corrected the algorithm "because the targeted and mass dissemination of hate and disinformation made the company a profit". These program routines should no longer remain a corporate black box, the parliamentarian demands: otherwise "we as a society and politicians will have no chance of countering them".

(nie)