Elon Musk Twitter acquisition effects – Science

Study Finds Community Notes Effectively Curb the Spread of Misinformation on X

Courtney Benton — Fri, 19 Sep 2025 21:20:55 +0000

In 2022, following Elon Musk’s acquisition of Twitter—now rebranded as X—the platform underwent significant changes that reshaped its approach to content moderation and misinformation control. Most notably, X dramatically reduced its content moderation team by 80%, pivoting instead to a crowdsourced fact-checking system known as Community Notes. Originally introduced as a pilot program on Twitter, Community Notes empowers everyday users to propose contextual annotations or corrections on specific posts suspected of containing inaccurate or misleading information. When these notes receive affirmative votes from a diverse group of users, as validated by X’s proprietary algorithm, the annotations become visible appended to the original posts. Since their implementation on X, Community Notes have garnered attention and adoption by other major social media platforms, including Meta and YouTube, reflecting a growing trend toward collaborative content verification across the digital landscape.

In a groundbreaking study led by researchers at the University of Washington’s Information School, the real-world impacts of Community Notes on X were rigorously analyzed. Tracking 40,000 posts that had notes proposed between March and June 2023, the team sought to understand how the presence of Community Notes influenced user engagement and the diffusion—or spread—of content through social networks. A crucial finding emerged: posts appended with Community Notes were significantly less likely to go viral. Specifically, within 48 hours of a note being added, reposts dropped by 46%, while likes decreased by 44%. These results suggest that Community Notes play a measurable role in mitigating the viral spread of potentially false information, signaling an important advancement in the fight against misinformation online.

Martin Saveski, the study’s senior author and an assistant professor at the University of Washington, emphasized the multifaceted nature of misinformation spread on social media platforms. According to Saveski, while Community Notes effectively reduce engagement metrics that commonly signal endorsement—such as reposts and likes—they represent only one tool within a broader strategy needed to combat the complex dynamics of false content dissemination. The team’s research, published in the prestigious journal Proceedings of the National Academy of Sciences in September 2025, provides empirical evidence supporting the value of Community Notes but also underscores the necessity of integrating diverse approaches for a comprehensive solution.

The study’s methodology involved a detailed comparative analysis between posts with helpful notes attached and those without. Out of the 40,000 posts with suggested notes, 6,757 were evaluated as helpful and subsequently displayed on the platform. The researchers monitored these posts for a 48-hour period following the appending of notes, assessing key indicators such as likes, reposts, replies, and views. The nuanced metrics paint a detailed picture: while reposts and likes saw the most substantial declines (46% and 44% respectively), replies fell by 22% and views by 14%. These figures highlight that engagement signaling active support or endorsement of the content diminishes more dramatically than passive consumption, such as viewing.

A critical aspect of misinformation’s virality lies in how widely and quickly content spreads within and beyond an account’s immediate follower network. Isaac Slaughter, lead author and UW doctoral student, explained that Community Notes altered the diffusion patterns of posts significantly. Posts flagged with notes saw reduced interaction from users distant in the social graph—those who do not follow the original poster or are several degrees removed. Conversely, users closer to the source, such as direct followers, exhibited a lesser degree of reduced engagement, implying that misinformation’s core support base may remain somewhat resilient despite fact-checking efforts. This finding points to the social network structure’s role in moderating the effectiveness of interventions like Community Notes.

Another enlightening revelation from the research was the differential impact of Community Notes based on content type and popularity. Notes added to posts containing altered media—such as doctored photos and videos—had a stronger suppressive effect on engagement than those appended to text-only posts. Additionally, Community Notes appended to highly popular posts showed even greater reductions in engagement and diffusion metrics. Timing also proved essential: notes that appeared promptly—within hours after posting—were markedly more effective. Delays nearing or exceeding 48 hours diminished the notes’ capacity to alter user behavior, reflecting the rapid lifecycle typical of viral misinformation.

Through examining these granular dynamics, Saveski’s research lab is now focusing on developing technological solutions to expedite the note attachment process. Accelerating the speed at which Community Notes appear could enhance their capacity to curb the spread of false information before it gains irreversible momentum. This work aligns with the broader imperative in social media governance to deploy real-time, scalable moderation tools that can operate effectively in fast-moving digital environments.

Despite these promising findings, the researchers also acknowledged limitations tied to data accessibility and platform policy changes. Since the study focused strictly on posts with notes proposed in early 2023, its scope does not encompass recent upgrades to Community Notes mechanisms or variations implemented on other social networks. Furthermore, X’s decision to end free access to its application programming interface (API) poses a significant obstacle for ongoing academic research, restricting independent scrutiny and longitudinal studies that are essential for understanding these moderation tools’ evolving efficacy.

The research team called attention to the broader challenges facing distributed moderation systems across platforms. Saveski raised pivotal questions about the sustainability of separate, siloed fact-checking communities operating independently on platforms such as X, TikTok, and Instagram. The motivation and bandwidth of users to contribute may wane if moderation efforts become fragmented. Additionally, cross-platform collaboration and data sharing could amplify the impact and scale of these systems, yet such cooperation remains limited. While X has made strides in opening its code and datasets, other major platforms have yet to commit to similar transparency, highlighting a critical bottleneck in efforts to collectively tackle misinformation.

At its core, this study provides a striking validation of Community Notes as a valuable asset in the arsenal against online misinformation, showcasing quantifiable reductions in user engagement with misleading content. The work by Saveski, Slaughter, and their colleagues—including co-authors Axel Peytavin from Stanford University and Johan Ugander from Yale University—represents a significant step forward in understanding how participatory moderation technologies shape information ecosystems. Funded in part by the University of Washington Information School’s Strategic Research Fund and the Army Research Office’s Multidisciplinary University Research Initiative, their findings pave the way for further innovation in real-time fact-checking and community-driven content governance.

As the digital information landscape continues to evolve, these insights could inform policymakers, platform designers, and the global public about the continuing effort needed to preserve information integrity. Community Notes may not be a panacea, but their demonstrated ability to curb engagement with misinformation highlights the promise of harnessing collective social intelligence to foster healthier online discourse.

Subject of Research: The effectiveness of Community Notes in reducing engagement and diffusion of false information on social media, specifically on X (formerly Twitter).

Article Title: Community notes reduce engagement with and diffusion of false information online

News Publication Date: 18-Sep-2025

Web References:

References:
Slaughter, I., Saveski, M., Peytavin, A., & Ugander, J. (2025). Community notes reduce engagement with and diffusion of false information online. Proceedings of the National Academy of Sciences, 10.1073/pnas.250341312

Keywords:
Social media, misinformation, content moderation, fact-checking, Community Notes, online virality, information diffusion, algorithmic governance, user engagement, altered media, digital communication, social network analysis

Try AI to Detect Toxic Speech Online: A New Science Magazine Insight

Courtney Benton — Thu, 05 Jun 2025 20:56:08 +0000

In recent years, social media platforms have become battlegrounds in the fight against toxic speech and online abuse, raising urgent questions about how to effectively moderate content at scale. Earlier this year, significant shifts occurred when Facebook announced a rollback of some of its rules against hate speech and abuse, a move that coincided with changes at X, the platform formerly known as Twitter, following its acquisition by Elon Musk. These developments have eradicated certain guardrails, making it increasingly difficult for users to avoid encountering offensive and toxic content as the balance between open discourse and user protection tilts precariously.

The daunting challenge of content moderation has long vexed social media companies, particularly due to the sheer volume of daily postings and the persistent resilience of toxic behavior online. Traditional moderation practices, which often involve human reviewers, come with serious drawbacks: human moderators face psychological trauma from constant exposure to hateful and violent material, while also struggling to keep pace with the flood of content. This context has catalyzed interest in using artificial intelligence (AI) as a technological intervention to assist in content screening, promising the ability to process vast quantities of data rapidly and with less human emotional toll.

Yet the adoption of AI in managing toxic speech is far from straightforward. Maria De-Arteaga, an assistant professor specializing in information, risk, and operations management at the University of Texas McCombs School of Business, stresses a critical dual challenge faced by AI systems: ensuring both accuracy and fairness. While an algorithm may perform admirably on average in detecting and flagging toxic language, its efficacy may be uneven across different social groups or contexts, resulting in disproportionate misclassification that can alienate small but significant communities.

De-Arteaga elucidates that it is possible for a model to display overall strong performance metrics, yet systematically err for particular groups — for instance, being sensitive in detecting speech offensive to one ethnic group but poorly recognizing similar toxicity aimed at another. This unevenness presents ethical and practical difficulties because biased moderation not only fails to protect those harmed by toxic speech, it can also inadvertently censor legitimate expression, undermining trust in platform governance.

To confront these limitations, De-Arteaga and her colleagues have advanced novel research that seeks to optimize AI models for both fairness and accuracy simultaneously. By developing a specialized algorithmic approach, their work enables platform designers and stakeholders to identify and navigate the trade-offs between these sometimes competing objectives, allowing for tailored moderation strategies that reflect the contextual needs and values of specific platforms and their user bases.

This research team, including Professor Matthew Lease, graduate students Soumyajit Gupta and Anubrata Das from UT’s School of Information, and Venelin Kovatchev from the University of Birmingham in the United Kingdom, utilized substantial datasets consisting of over 114,000 social media posts pre-classified as “toxic” or “nontoxic” by previous studies. These datasets provided a rich ground truth benchmark for training and evaluating machine learning models designed for toxicity detection across diverse online communities.

Central to their methodology is the use of a fairness metric known as Group Accuracy Parity (GAP), which measures the balance of model accuracy across distinct demographic groups rather than merely considering an aggregate score. Applying GAP within their algorithmic framework, the researchers integrated fairness constraints directly into the model training process, fundamentally shifting AI content moderation from a performance-only focus to one that respects equity and social justice considerations as first-class objectives.

The results are compelling: their approach not only surpassed the next-best solutions by up to 1.5% in fairness metrics but also achieved superior outcomes in simultaneously maximizing both fairness and accuracy. This advancement illustrates an important breakthrough in AI ethics and machine learning, offering a practical pathway for developers and platform administrators to mitigate bias while maintaining robust detection capabilities against toxic speech.

Despite this promising progress, De-Arteaga cautions that GAP and similar fairness measurements are not universal remedies. Different platforms and policymakers may subscribe to varying definitions of what constitutes fairness, influenced by cultural, social, and political contexts. Moreover, concepts of toxicity and abuse are inherently fluid, evolving over time as societal norms shift, which challenges the static nature of algorithmic systems trained on historical data.

Getting these nuances right is more than an academic exercise — mislabeling a user’s speech as toxic when it is not can unjustly exclude individuals from vital public conversations, damaging their digital presence and voice. Conversely, failing to identify and act on harmful content exposes users to real risks of harassment, psychological harm, and the erosion of online community standards. For global platforms like Facebook and X that operate across varied jurisdictions and serve heterogeneous user groups, the stakes are immensely high.

Addressing these complexities requires a multifaceted approach. De-Arteaga emphasizes the importance of designing AI that is adaptable, capable of being updated constantly to reflect shifting societal understandings and regional sensitivities. This design philosophy calls for the thoughtful collection and curation of training data that encapsulate diverse perspectives and contexts, ensuring the AI’s relevance beyond singular national or cultural paradigms.

Transparency remains a cornerstone of this endeavor. By making the GAP algorithm’s code publicly available, the researchers invite broader scrutiny, collaboration, and refinement from the global scientific and tech communities. This open-source model facilitates ongoing improvement and encourages platforms to implement context-aware, justice-oriented solutions rather than one-size-fits-all fixes.

Ultimately, the intersection of technology and social values demands interdisciplinary engagement. Achieving effective and fair detection of toxic speech goes beyond mere algorithmic prowess; it mandates expertise in social sciences, linguistics, ethics, and user behavior. De-Arteaga encapsulates this need, emphasizing that “you need to care, and you need to have knowledge that is interdisciplinary,” underscoring the vital human dimension behind technological innovation.

The breakthrough outlined in this research marks a significant milestone in the quest for responsible and equitable AI moderation tools. By navigating the Pareto trade-offs — the inherent balancing act between fairness and accuracy — the study opens doors to more nuanced and just approaches to combatting toxic speech online while safeguarding the pluralistic nature of digital discourse.

As online platforms wrestle with the dual imperatives to protect users from harm and preserve open speech, such advances in AI moderation represent a critical step forward, informing future policies and technological frameworks that uphold dignity and fairness at scale in the digital public square.

Subject of Research: Fairness and accuracy in AI-based detection of toxic speech on social media platforms.

Article Title: Finding Pareto trade-offs in fair and accurate detection of toxic speech

News Publication Date: 11-Mar-2025

Web References:

References: Publicera KB – Information Research, 2025.

Keywords: Social media, Mass media, Communications, Media violence, Propaganda, Marketing, Written communication, Linguistics