Tamara Yesmin Toma

Researcher, Dismislab
YouTube Removes 150,000 Videos from Bangladesh in Three Months Over Policy Violations

YouTube Removes 150,000 Videos from Bangladesh in Three Months Over Policy Violations

Tamara Yesmin Toma

Researcher, Dismislab

YouTube has removed over 150,000 videos uploaded from Bangladesh within a span of three months, citing non-compliance with community guidelines. According to Google’s transparency report, videos were removed during the months of October to December last year. Globally, the platform eliminated more than 9 million videos during this quarter. Among the 30 countries listed, India topped the chart for the highest number of video removals, while Bangladesh is in the eighth position.

Videos have been removed for various reasons, including spreading harmful content that may incite violence, posing safety risks to children, containing violent or insensitive scenes, promoting nudity and sexual content, and spreading false information. Additionally, over 20 million YouTube channels were also deleted from the platform during the same period.

The most common type of video removed for violating community guidelines was harmful or dangerous content, accounting for 39.2 percent of removals. This includes content deemed to pose a risk of inciting violence or imminent harm. Following closely behind, videos concerning child safety policy comprised 32.4 percent of removals, with the aim of preventing harassment and online abuse directed at minors. The third most removed category was violent or graphic content, constituting 7.5 percent of removals. YouTube defines violent or graphic content as footage depicting events such as riots or scenes of bloodshed.

YouTube verifies policy violations through a combination of human review and machine learning algorithms, specifically artificial intelligence (AI). Remarkably, 96 percent of the removed videos were identified and flagged by AI. The remaining 4 percent were flagged based on complaints or suggestions from YouTube’s general users and members of the ‘Priority Flagger Program‘. This program, comprising both government agencies and NGOs, actively monitors violations of YouTube’s Community Guidelines.

YouTube stated that 53 percent of removed videos from the fourth quarter of 2023 were removed without being viewed. Additionally, a significant 27 percent of videos received only one to 10 views before removal.

Alongside video removals for violations of community guidelines, YouTube also removed 110 million comments. Remarkably, approximately 99 percent of these comments were automatically flagged as spam and subsequently deleted.

Out of the 9 million videos removed in the last three months of the year, the majority, up to 2.25 million, were removed from India. Following India, Singapore and the US ranked second and third respectively in terms of video removals. Singapore accounted for 12.4 million removed videos, while the United States had 7.8 million removed videos. Indonesia ranked fourth, with over 770,000 videos removed.

YouTube Removes 638,000 Videos from Bangladesh in One Year

“According to the Google Transparency Report, total 638,440 videos uploaded from Bangladesh were removed in 2023. In the last three months of 2022, the platform removed 115,907 videos from Bangladesh. This indicates a 31.19 percent increase in the video removal rate during the last three months of 2023 compared to the same period in 2022.

Dismislab recently published a report on the burgeoning market of cheap fakes that has emerged on YouTube, particularly targeting Bangladeshi viewers. A case study of three channels monetizing political disinformation on YouTube reveals that, following the dissemination of such content, these channels experienced an increase in both views and subscribers. Furthermore, it is noted that YouTube profits from advertising on this disinformation content.

Earlier, Dismislab published another report titled “YouTube and YouTubers make a killing with fake death videos!” on how YouTube is profiting from celebrity fake news. This report sheds light on how YouTube profits from celebrity fake news. According to the findings, YouTube channels capitalize on people’s fascination with stars to earn money through advertising. Additionally, YouTube itself receives a share of the revenue generated from this misinformation business.

“YouTube Guidelines Consistently Enforced Worldwide”

Since 2006, YouTube has been operated as a company owned by Google. The company emphasised its commitment to safeguarding the YouTube community from harmful content since its early days. “YouTube’s Community Guidelines have protected the YouTube community from harmful content since the earliest days as a company,” the statement said. It further asserted that these guidelines are consistently enforced worldwide, regardless of the uploader, content origin, or generation method.

YouTube also revealed significant enforcement actions in its statement. “YouTube removed over 20 million channels in Q4 2023 for violating our Community Guidelines (globally),” the statement disclosed. It specified that the majority of these channels were terminated due to violations of spam policies, including scams, misleading metadata or thumbnails, and spam in videos and comments.

Meta too has taken steps to remove harmful coordinated networks

Meta recently released its quarterly Adversarial Threat Report for Q4 2023, emphasising its commitment to implementing Community Standards and combating various harmful activities like cyber espionage and influence operations. The report provides a comprehensive overview of Meta’s ongoing efforts to counter adversarial behaviours across its platforms.

A significant highlight of the report is Meta’s endeavours against the surveillance-for-hire industry. Insights from Meta’s investigations into eight firms from Italy, Spain, and the United Arab Emirates reveal operations targeting individuals worldwide through various malware aimed at iOS, Android, and Windows devices. These malicious programs possess capabilities to gather sensitive information such as device data, location, photos, contacts, and communication activities on social media and messaging apps. Meta’s post-disruption analysis suggests that its actions have significantly impeded spyware groups, resulting in a reduction in their activities on Meta’s apps.

Additionally, the report offers updates on Meta’s takedowns of new CIB (Coordinated Inauthentic Behaviour) networks in China, Myanmar, and Ukraine. In China, Meta dismantled a network targeting US audiences, masquerading as members of US military families and anti-war activists across various internet services. In Myanmar, Meta thwarted a network targeting domestic audiences, presenting as members of ethnic minorities and linked to individuals associated with the Myanmar military. In Ukraine, Meta eliminated a network targeting audiences in Ukraine and Kazakhstan, identified through information shared by Google.

Moreover, the report underscores Meta’s efforts to counter domain name abuse globally. Meta resolved a legal dispute against Freenom, a country code domain registry provider, whose domain names accounted for over half of all phishing attacks involving country code top-level domains (ccTLDs).

Additionally, the report includes an annual update on adversarial trends identified since Russia’s full-scale war against Ukraine, spotlighting Meta’s actions against Russian state-controlled media and covert influence operations. Meta observed a decline in posting volumes and engagement levels by Russian-origin CIB campaigns, signifying the efficacy of its enforcement measures.