On a humid October afternoon in Colombo, a roomful of election officials, journalists, and civil society activists gathered not to debate poll numbers or campaign strategies, but to confront something far more elusive and corrosive: the quiet, often invisible manipulation of the digital spaces where democracy now lives. The occasion was “Beyond the Algorithm,” a forum organized by Democracy Reporting International (DRI), held on 30 October, designed to wrestle with a question that has become urgent in Sri Lanka and across the world—how do social media platforms, artificial intelligence, and the hidden logic of algorithms shape what voters see, believe, and ultimately do at the ballot box? The timing was no accident. Sri Lanka has just passed through an intense 2024–2025 election cycle, one that included a presidential election, a parliamentary election, and local government polls. Each of these contests unfolded against a backdrop of rapidly evolving digital threats: fake accounts, manipulated images, viral falsehoods, and a disturbing wave of online violence targeting women in politics. The forum brought together representatives from the Election Commission of Sri Lanka (ECSL), media organizations, and civil society to unpack these dangers and search for ways to protect the integrity of election-related information without smothering the free expression that democracy itself depends upon. At the heart of the gathering was a simple but profound realization: the battle for democracy is no longer fought only in polling stations or parliamentary chambers, but in the endless scroll of social media feeds, where truth and falsehood compete on unequal footing, and where a single fabricated video can do more damage than a thousand speeches.
The research presented at the forum painted a stark picture of just how deep the problem goes. Camila Weinmann, DRI’s Digital Democracy Research Officer, and Prihesh Ratnayake, a Digital Research Analyst, shared findings from an examination of Sri Lanka’s recently concluded elections, and the evidence was sobering. During the election cycle, researchers identified sixty-six TikTok accounts that were impersonating political parties and candidates. These accounts did not merely exist in the shadows; 45.5 percent of them were active during the 2024 Presidential Election, and 21.2 percent remained active all the way into the 2025 Local Government Elections. Impersonation of this kind is particularly insidious because it weaponizes trust. A voter might see a policy statement, a scandalous accusation, or an official-looking announcement, and believe it came directly from a party or candidate, when in reality it was fabricated by an unknown actor with unknown motives. The research team reported the offending accounts to TikTok and the ECSL, and fifty-one of them were eventually removed. But the persistence of the others underscores a broader vulnerability. The research also revealed that women politicians faced higher levels of technology-facilitated gender-based violence during the same period—a chilling reminder that digital manipulation is not merely a matter of misleading information, but also of silencing and intimidating those who seek to participate in public life. The study went further, examining the spread of inaccurate electoral information through chatbots, the continued publication of political advertisements during the legally mandated silence period of the parliamentary election campaign, and the emerging risks posed by generative artificial intelligence, which can manufacture convincing falsehoods at scale and with terrifying speed.
Behind these findings lies a deeper concern that DRI Executive Director Nicola Schmidt articulated with quiet urgency. Opening the event, she reminded the audience that disinformation is not a uniquely Sri Lankan problem, but a global phenomenon. Generative AI, deepfakes, and the algorithmic architecture of social media platforms have created an environment in which misleading information can travel across the world in seconds, often faster than the corrections that follow it. “Social media and different platforms are contributing to the speed of the spread of disinformation,” Schmidt said. “We as DRI are looking into this globally, looking at what algorithms do.” Her words carried added weight because the forum’s title, “Beyond the Algorithm,” was a deliberate challenge. It asked participants to look past the surface of the content they consume and toward the hidden machinery that decides what goes viral and what gets buried. Algorithms are not neutral; they are designed to maximize engagement, and engagement thrives on outrage, fear, and novelty—emotions that falsehoods often exploit with devastating effectiveness. Schmidt stressed that understanding this algorithmic contribution to disinformation is essential to promoting digital rights and holding technology platforms accountable for the consequences of their designs. In that sense, the struggle against disinformation is not merely a technical problem to be solved with better fact-checking tools, but a question of responsibility. Platforms cannot continue to claim that they are simply neutral conduits while their algorithms actively amplify harmful content. The call for accountability was clear, and it resonated throughout the day’s discussions.
That call was echoed and expanded by A M Faaiz, a member of the Election Commission of Sri Lanka, who offered a reflection that grounded the conversation in the human realities behind the screens. “Algorithms decide what appears on our screen, but algorithms do not vote,” he observed. “Behind every screen sits a citizen and behind every ballot lies a judgment that a citizen has formed from what he or she has read, heard and come to believe.” It was a powerful reminder that the ultimate goal of electoral integrity is not merely clean databases or secure voting machines, but the quality of the decisions that citizens make. If those decisions are formed in an environment saturated with deception, manipulation, and abuse, then the legitimacy of the entire democratic process is called into question. Faaiz went on to reveal that the ECSL, with support from DRI and Factum, was taking concrete steps to respond. The Commission was formulating a strategic action plan on disinformation and harmful content, and mapping its digital capacities to develop a roadmap for its own digital presence. “What began as a project is becoming part of the institution,” he said. His words suggested a welcome shift: rather than treating digital threats as an ad hoc challenge to be managed during election seasons, the Commission was working to institutionalize resilience, building permanent structures that could anticipate and respond to manipulation year-round. But Faaiz was careful to sound a note of caution as well. He stressed that measures to safeguard electoral integrity must remain consistent with freedom of expression and media freedom, warning that such efforts should never be used to suppress legitimate political debate or criticism—including criticism of the Election Commission itself. “Our purpose is to protect the voters, not to police opinion,” he said, drawing a bright line between safeguarding the democratic process and silencing dissent.
The first panel discussion of the day, titled “The Election Information Ecosystem,” dove deeper into the practical challenges of navigating this treacherous landscape. Panellists examined the protocols that traditional newsrooms could adopt when public figures make unverified claims, a situation that has become increasingly common in the era of social media, where a single statement from a prominent politician can go viral before journalists have had any chance to check its factual basis. The discussion also explored the effectiveness of counter-narratives delivered through drama, podcasts, and interviews—an acknowledgment that simply debunking falsehoods is often not enough. People are moved by stories, not spreadsheets, and efforts to build resilience against disinformation need to engage citizens emotionally and culturally, not just intellectually. The conversation turned to the effectiveness of voluntary codes of conduct and memorandums of understanding between the ECSL and technology platforms. Such agreements are often seen as a softer alternative to legislation, but they have a mixed record, depending on the willingness of platforms to honor their commitments. The panellists also highlighted the need for stronger legislative support, while wrestling with the tension between regulation and freedom. Perhaps most worryingly, they pointed to growing political polarisation as a fundamental threat to information integrity. In a deeply divided society, citizens are more likely to accept information that confirms their existing beliefs and reject anything that challenges them, no matter how well-sourced. Polarisation creates fertile ground for disinformation, because it erodes the shared standards of evidence and truth that make democratic deliberation possible.
The second panel, titled “Resilience in Practice,” shifted the focus from diagnosis to remedy, drawing on hard-won experiences from fact-checking, advocacy, and media literacy initiatives. Participants spoke candidly about the difficulties of verifying information in real time, particularly in a country where multiple languages and complex contexts make on-the-ground confirmation challenging. They discussed the often frustrating responses from social media companies during fact-checking processes, where takedown requests can languish in limbo while falsehoods continue to spread, and where transparency about decision-making remains sorely lacking. There was also a recognition of the vital importance of equipping local journalists to counter disinformation, not only with digital tools but with the editorial judgment and institutional support they need to do their work under pressure. The conversation did not shy away from the uncomfortable reality that fact-checking initiatives consistently struggle to attract funding. The work is labor-intensive, often thankless, and does not produce the kind of viral moments that attract donors. Yet it is precisely this slow, unglamorous labor that forms the backbone of democratic resilience. DRI Project Manager Aingkaran Kugathasan closed the event by thanking participants and partners for their contributions, but the sense of urgency lingered long after the formal proceedings ended. The forum was held under the “Strengthening Resilience Against Disinformation in Sri Lanka” project, co-funded by the European Union, and it brought together international partner organizations, stakeholder institutions, media representatives, and civil society groups in a collective effort to defend the integrity of democratic life. The challenges are immense, and the algorithms will only become more sophisticated, but the day offered a glimpse of what resilience looks like: not a single dramatic solution, but a patient, collaborative effort to ensure that when citizens cast their ballots, they do so on the basis of information they can trust—and in a digital world that no longer silences the voices of women and all those who dare to participate in public life.