Дижитал мартагнуул: ОХУ болон авторитар дэглэмүүд түүхийг хэрхэн засварлаж байна вэ?

The disappearance of independent journalism, firsthand accounts and historical archives from the internet is helping Russia and other authoritarian states rewrite history and shape collective memory, researchers warn.

From altered reporting on the Beslan school siege to vanishing articles documenting anti-government protests, the loss of online records — a process that researchers call “digital oblivion” — is contributing to public “amnesia” about key events that autocrats find unhelpful.

As artificial intelligence becomes a gateway to news and information, the researchers warn that gaps in the historical record combined with a flood of propaganda can shape the answers people receive about the past.

“Digital tools are used to consolidate what would have otherwise been a more painstaking process,” Steven Feldstein, a researcher with the Carnegie Endowment for International Peace, said at a recent panel at the Craig Newmark Graduate School of Journalism at the City University of New York.

Autocratic leaders who have long relied on violence, intimidation and legal harassment to consolidate power have added powerful digital tactics to their toolkit. These can make independent accounts of key events harder to find, unsafe to publish or impossible to verify.

Digital oblivion can involve a government’s or a company’s deliberate removal of a website or archive, or can be the quiet disappearance or blocking of online photos, blogs or social media posts. Records can also become inaccessible through broken links and obsolete media formats, even when the cause is unclear.

The erasure can be ad hoc or systematic. It is more expansive than old-school censorship.

“Contemporary autocratic politics is all about a ‘glorious past.’ The thing that needs to happen is the ‘past’ needs to be cleaned up,” said panelist M. Gessen, co-founder of Kronika, an initiative to preserve independent journalism.

Documentation of large-scale anti-Putin protests in Moscow in 2015 has become difficult to find online, as has material tracing the development of Russia’s LGBTQ+ movement, Gessen said.

In some cases, reporting remains available while passages challenging the official account disappear.

A Russian translation of “The School,” journalist C.J. Chivers’ account of the 2004 Beslan hostage crisis for Esquire, was heavily edited around the tragedy’s 20th anniversary, according to a 2025 analysis by Mediazona using the Wayback Machine.

The translation had been preserved on the website of Pravila Zhizni, the magazine that replaced Russian Esquire after it lost its license in 2022.

Mediazona’s analysis found that parts of the online article changed, including a passage describing how most of the deaths occurred after Russian troops stormed the school. The article’s conclusion that the siege had shaken faith in the Russian government and its ability to protect the Russian people also disappeared.

An investigation by Citizen Lab, a Canada-based research group, documented extensive blocking of content on Ukraine, Belarus and LGBTQ+ issues, as well as accounts belonging to independent media and users opposing Putin or supporting Ukraine, on Russian social network VKontakte.

At times, ordinary people have voluntarily become tools of digital oblivion, the panelists said.

Gessen cited the example of a Belarusian woman who deleted photos documenting her participation in anti-government protests following President Alexander Lukashenko’s disputed 2020 re-election in order to protect herself.

China’s restrictions on information about the June 4, 1989, Tiananmen massacre mark an early success story for digital oblivion. Online censorship makes accounts of the military’s crackdown on the student-led protests difficult to find on the Chinese internet, while documentation of subsequent protests in Hong Kong has also been heavily restricted.

The panelists said similar pressures on the public record can emerge in democracies.

Gessen recalled the case of a U.S. scientist who removed language about climate change from existing research and inserted language acceptable to the Trump administration to safeguard funding from the National Institutes of Health.

Panelists also pointed to the Trump administration’s removal of information from government websites and changes that have made studies and reports harder to find online.

AI is accelerating the impact of these disappearances, the researchers warned. Digital oblivion regularly leads to large language models being trained on faulty and incomplete sources, or censored information. Decisions by AI companies about which material to use and how to present it can then influence what people know and understand about a historical event.

“AI companies have become increasingly opaque in how they are training their models,” Sarah Myers West, co-director of the AI Now Institute and an expert on technology policymaking, said at the panel, titled “Digital Oblivion: How Autocrats Engineer Forgetting.”

A flood of low-quality, AI-generated content often called “AI slop” can create further noise, making reliable accounts harder to find.

Pro-Kremlin information operations have demonstrated how large volumes of misleading content can influence AI responses.

NewsGuard, a company that analyzes misinformation, said in a March 2025 report that the pro-Kremlin disinformation network Pravda was flooding search results and material retrieved by AI chatbots with false claims and propaganda.

A separate NewsGuard analysis showed a video produced by Russian influence operation Storm-1516 was provided by an AI chatbot as evidence that Ukrainian soldiers had burned an effigy of U.S. President Donald Trump and promoted other false claims about Ukrainian President Volodymyr Zelensky.

A November 2025 analysis by the Institute for Strategic Dialogue identified roughly 900 political websites that have linked to Pravda network articles, with the majority citing them as credible sources.

The institute said the network had published more than 6 million articles since its launch in 2014 and was publishing new content every few seconds.

Efforts to preserve that record are growing, such as in librarians’ work to save data and archives, citizens protesting data-center construction and communities rejecting Flock cameras.

In Russia, activists are working to safeguard the history of the LGBTQ+ movement as state restrictions threaten both its participants and its archives.

Readers can also reduce their exposure to manipulated narratives by consulting original reporting and verifying information through multiple reliable sources, rather than through the framework, and lack of context, of an AI platform, panelists said.

But preserving access to reliable information also requires action from governments as AI infrastructure becomes increasingly embedded in institutions, schools and systems, they argued.

“They need to have real incentives to train their models on facts,” said Gessen.