Seznam

Report March 2025

Submitted
Seznam.cz is a leading Czech internet company that provides various online services, such as search, news, email, maps, and e-commerce. Seznam.cz is proud to be a signatory of the EU Code of Practice on Disinformation and to contribute to its mission of enhancing transparency, accountability, and trustworthiness of online platforms and information. We believe that our actions and achievements in 2024 demonstrate our commitment and our progress in this regard.
We are aware that the fight against disinformation and misinformation is an ongoing and evolving challenge that requires constant vigilance, adaptation, and cooperation. We are determined to continue our efforts and cooperation with other signatories, stakeholders, and authorities to uphold the Code’s principles and objectives.
This report covers the following aspects of our approach to misinformation and disinformation on our services: Seznam Search, Seznam Newsfeed, and Discussions.

Our commitment to fight falsehoods
- We take the issue of disinformation and misinformation very seriously. Since 2023, we have a fulltime role of anti-disinformation officer who coordinates all projects across services that we aim to keep trustworthy, clean of falsehoods and manipulations.
- We have strict policies tackling impermissible manipulative behaviours and practices on our services, including the policies countering manipulative practices for AI. Our long-term goal is to promote transparency and accountability to our users and partners regarding these policies and actions.
- We combine human editors and technologies to implement and enforce these policies, elevate authoritative information and high-quality content, promote good journalism and moderate discussions.
- We educate users of our services by explanatory labels, fact-checks, links to credible sources, promotion of quality journalism and prioritization of quality content in our ranking systems. 
- We also provide features to easily report problematic content and personalize user experience, thus empowering them to get trustworthy information and act upon them.
- We develop projects that increase media literacy and critical thinking skills among people of all ages. 

Our achievements and improvements
- We have continued to improve our detection of AI-generated content, clickbait articles, and other forms of misleading or harmful content, and we have increased our human oversight and moderation.
- We continuously trained our AI system to detect misinformation narratives in comments and suggest relevant fact-checks to editors.
- We have measured the impact of our fact-checking, proving that it increases perceived trustworthiness and objectivity among the public, as well as it decreases stress and negative emotions. 
- We have made all our fact-checks public for anyone to double-check it or use it.
- We have continued to blacklist and demote misinformative and poor-quality sources on our services, while using independent external ratings. Our steadfast partnership with impartial civic and academic institutions ensures a fair and objective rating of the media outlets we aggregate, rank, and display within our services and sites.
- We have continued to flag misinformation websites and display warnings in Seznam Search.
- On our blogging service, Seznam Medium, we have continued to keep the same editorial standards and commitments to prevent the spread of misinformation.

Commitment to the Code and its Signatories
- We have played an active role in different Subgroups of the Code, particularly on Monitoring & Reporting and AI. We are open to share our data, feedback, and best practices with other Signatories and the European Commission, to improve the implementation and evaluation of the Code.
- We want to keep sharing our learnings and insights with other Signatories, as well as with the public and the media. We believe that transparency and dialogue are essential for building trust and credibility in the online space.

We hope that this executive summary provides a clear and comprehensive overview of our actions and achievements in 2024 to fight against disinformation and misinformation on our services. We are proud of our progress, and we are determined to continue our efforts and cooperation with other signatories, stakeholders, and authorities to uphold the Code’s principles and objectives.

Download PDF

Commitment 14
In order to limit impermissible manipulative behaviours and practices across their services, Relevant Signatories commit to put in place or further bolster policies to address both misinformation and disinformation across their services, and to agree on a cross-service understanding of manipulative behaviours, actors and practices not permitted on their services. Such behaviours and practices include: The creation and use of fake accounts, account takeovers and bot-driven amplification, Hack-and-leak operations, Impersonation, Malicious deep fakes, The purchase of fake engagements, Non-transparent paid messages or promotion by influencers, The creation and use of accounts that participate in coordinated inauthentic behaviour, User conduct aimed at artificially amplifying the reach or perceived public support for disinformation.
We signed up to the following measures of this commitment
Measure 14.1 Measure 14.2
In line with this commitment, did you deploy new implementation measures (e.g. changes to your terms of service, new tools, new policies, etc)?
If yes, list these implementation measures here
Do you plan to put further implementation measures in place in the next 6 months to substantially improve the maturity of the implementation of this commitment?
If yes, which further implementation measures do you plan to put in place in the next 6 months?
Measure 14.1
Relevant Signatories will adopt, reinforce and implement clear policies regarding impermissible manipulative behaviours and practices on their services, based on the latest evidence on the conducts and tactics, techniques and procedures (TTPs) employed by malicious actors, such as the AMITT Disinformation Tactics, Techniques and Procedures Framework.
QRE 14.1.1
Relevant Signatories will list relevant policies and clarify how they relate to the threats mentioned above as well as to other Disinformation threats.
Seznam Search 
·       Our search systems prioritize high-quality and authentic information sources to provide users with the most relevant content and best user experience. At the same time, we reduce the reach of low-quality content, often spread through deceptive or manipulative behavior listed in Commitment 14, by spam detection tools. 
·       While these measures address TTPS relevant for search engines (e.g. inauthentic domains or Inauthentic coordination of content creation or amplification) at a high volume, we also warn users against known misinformation websites. Warnings flagging untrustworthy content appear to users in the organic search results under the link to misinformation sites, and we do not display images and videos from these domains.

 
Seznam Newsfeed has policies for what can appear in the service to promote quality journalism and trustworthy news. 
·       Independent Media Rating published by the Foundation for Independent Journalism (NFNZ) is mandatory for all news websites displayed in Seznam Newsfeed. 
·       Websites that disseminate or contribute to the dissemination of misinformation or conspiracy theories are excluded from the feed. 
·       We also have a policy to limit clickbait headlines and other deceptive or manipulative editorial practices.
·        Furthermore, we provide Guidelines for our partners and require compliance with standards for information sourcing on all websites displayed in Seznam Newsfeed. 
·       To avoid deceptive and manipulative media, we insist that the reader must be able to trace the source of the information and make judgment on its credibility. There are even stricter rules for news and articles on health and scientific topics.


Seznam Discussions 
·       Our policies require users to appear under their real identity to address inauthentic and deceptive behaviors listed in Commitment 14. Fake accounts are not allowed at the service, and we have several levels of verification that apply proportionally to the behavior and reach of a particular account. We further limit deceptive practices by prohibiting content that impersonates any physical person or organization and engaging in inauthentic or coordinated behavior to deceive or mislead.
·       Any content in breach of both private and public law is forbidden on the service. We pay special attention to content that incites hatred towards individuals or groups of the population based on racial or religious affiliation and nationality. On top of that, vulgar or obscene language or vulgarity by description, slanders, and threats of violence are prohibited as well.
·       We also limit the spread of spam and other TTPs listed in Commitment 14. by targeting content with specific dubious features (e.g. all caps, special characters, etc.), re-posted content, and content that spreads misinformation or contains links to such websites. 
·       On top of that, we fact-check manipulative comments, label them and add context boxes instead of just simply deleting the post or banning the account.
·       Finally, we aim to prevent people from abusing our services for commercial or any other gain, so we remove content that shares personally identifiable information or other private information obtained from illegal sources or published without consent.
·       We do this by a combination of AI tools and human editors.


Seznam Search 
·       We use both machine learning and humans to prioritize high-quality and authentic content and reduce the reach of low-quality content, often spread through deceptive or manipulative behavior listed in Commitment 14. More information on how Seznam search works can be found here. 
·       Seznam Search also uses various external sources (EU Guidelines, National Center for Cybernetic Operations, the Foundation for Independent Journalism, the Atlas of Conspiracies project, and the Slovak initiative Konspiratori.sk) to identify and flag disinformation websites and display appropriate warnings.


Seznam Newsfeed uses a combination of humans and AI to evaluate quality of content.
·       Every day, around 3,000 media articles are submitted to the Newsfeed service by partner websites. About 2,000 of these texts are published after being automatically checked and pre-categorized using tools such as clickbait headline detection or topic specificity classifier. Additionally, almost half of the submitted texts undergo a manual review process by trained editors.
·       We provide our editors with detailed internal guidelines and use independent MediaRating published by the Endowment Fund for Independent Journalism (NFNZ) to evaluate submitted articles. Websites with the MediaRating B- and C are excluded from the Newsfeed. We also exclude websites that are listed by NFNZ and Konspiratori.sk as misinformative.
·       Only quality journalism is published automatically, but with editorial oversight. Based on the common standards of quality journalism, we have introduced a stricter obligation to source information and quotations properly even in texts that do not qualify as news (including blog posts). 
·       Only news outlets with the highest quality ratings are allowed to publish their contents directly in our Newsfeed. However, their articles are also checked retroactively by automated tools and human editors.

 
Seznam Medium
·       Our blogging service applies similarly strict editorial principles as Newsfeed does. 
·       According to our terms of service, bloggers must not disseminate or promote misinformation or conspiracy theories; may not link to misinformation websites (based on the lists of NFNZ, Nelež and Konšpirátori.sk), or use content of such websites. 
·       All blogs posted to Seznam Medium are checked by human editors who may unpublish blogs or ban users breaching the terms. We also have internal editorial guidelines that are followed.

 
Seznam Discussions, to combat TTPs, relies on AI and human editors assessing content, as well as fact-checkers moderating discussions.
·       We equipped discussion editors and fact-checkers with comprehensive internal guidelines. Our fact-checkers add context to manipulative or misinformative comments based on independent third-party sources such as EUvsDisinfo, AFP and other international and national trusted sources. 
·       Our editors verify and eventually ban fake accounts, and accounts that deploy TTPs or otherwise violate our policy.
·       However, our approach is focused on the behavior evaluation of each actor. We look for specific patterns of behaviors exhibited by malicious actors, and not only for violating content. Whenever possible, we prefer to correct the information and lead users with good intentions to credible sources of information. On the other hand, we pay special attention to the most common misinformation narratives and TTPs deployed by malicious actors and apply stricter sanctions.  
·       E.g. Following the onset of the war in Ukraine, we began banning accounts that consistently disseminated pro-Kremlin narratives, hate speech against Ukrainians, and support or justification for the Russian invasion. 
·       While most of the misinformation we currently address relates to the war in Ukraine, we also tackle myths and hoaxes about global security, vaccines, Czech politics, LGBTQI+ and others.