From Kenya To The US Midterms: Inside India’s Shadow Social Media Campaigns

A day before Americans vote in the November 3 midterm elections, an investigation by Al Jazeera has drawn attention to the growing international reach of Indian political consultancies and their alleged role in covert digital campaigning.

The investigation found that 28 accounts on X shared videos supporting US Democratic politician Haley Stevens ahead of the August 4 Michigan Senate primary, generating more than seven million views. At least 15 of the accounts appeared to be posting from India.

Stevens lost the Democratic primary to Abdul El-Sayed. The activity preceded the November midterms, but the investigation does not establish that her campaign commissioned the accounts, paid for their work or knew about their involvement.

The report traces a similar model to Kenya’s 2022 presidential election, where an Ahmedabad-based content creator said he produced material connected to William Ruto’s campaign. Together, the cases raise questions about how political messaging is outsourced, distributed through concealed networks and made to appear like spontaneous online support.

How Does A Shadow Social Media Campaign Work?

Political consultancies typically advise candidates on messaging, publicity and voter outreach. The reported shadow-campaign model takes some of that work outside a candidate’s visible communication channels.

The process can begin with a political brief from a client or an intermediary. Content creators then prepare posters, short videos, edited speeches and captions to promote a candidate or undermine an opponent. The material is circulated through social media pages and accounts that do not openly identify themselves as part of the campaign.

In the Kenya case, a 22-year-old creator identified as Rajan Kumar said he produced 50 social media posts for about $7 a day for work connected to Ruto’s 2022 presidential campaign. He described editing opponents’ speeches and making posters that promoted Ruto’s prospects. The content was shared through shadow accounts rather than the candidate’s official social media presence.

Distribution can involve existing pages, newly created accounts, rebranded pages or accounts available for hire. Multiple accounts may then circulate similar material, share one another’s posts or generate engagement to increase its visibility.

Paid influencers and content creators can also be part of this wider ecosystem. A real person publishing sponsored political material is different from an automated or fake account, although undisclosed payments can conceal the relationship behind a post. The Al Jazeera investigation documents paid content production in the Kenya example, but does not establish a specific paid-influencer arrangement in the Michigan case.

The purpose of using separate pages and intermediaries is to create distance between the political message and the candidate or organisation it supports. Kumar told Al Jazeera that he never worked directly with Ruto’s representatives and that creators were not necessarily told who was involved further up the chain.

What Do The Kenya And US Cases Show?

Kenya provides an earlier reported example of Indian content-production capacity being used in a foreign election. The official Independent Electoral and Boundaries Commission (IEBC) result shows that Ruto won the 2022 presidential election with 50.49% of the vote, against 48.85% for Raila Odinga.

The election result, however, does not establish who commissioned the online content or whether it influenced voting. Kumar’s account is testimony about work connected to the campaign, not proof that Ruto or his official campaign organisation directed the activity.

The Michigan case brings the reported model into the US election cycle. Al Jazeera cited the research group Antibot4Navalny, which identified 28 X accounts sharing videos supporting Stevens. Researchers at Clemson University’s Media Forensic Hub also identified similar tactics involving accounts and pages supporting her campaign.

But the evidence has limits. The identities of those controlling every account, the source of payments and the identity of any client or intermediary have not been established publicly. Nor is there evidence in the investigation that Stevens or her campaign knew about the network.

The distinction matters: content supporting a candidate does not, by itself, establish that the candidate authorised or benefited from a covert operation.

How Do Researchers Identify Coordinated Accounts?

Researchers look for patterns across accounts rather than treating an individual post or anonymous profile as proof of a campaign.

Repeated or near-simultaneous uploads, identical captions, shared videos and recurring hashtags can indicate coordination. Account histories may reveal sudden changes in names, profile pictures, subject matter or posting behaviour. Researchers also examine clusters of accounts that repeatedly like, repost or amplify one another’s material.

Location indicators and posting patterns can offer further clues. In the Michigan case, at least 15 of the identified accounts appeared to be posting from India. But an apparent posting location does not establish the operator’s nationality, identity or employer. Virtual private networks and changes to account settings can also complicate efforts to establish where an account is being operated.

Advertising records provide another source of evidence. Meta’s Ad Library allows researchers to examine political and issue-based advertisements, including the associated page, creative material, delivery dates and ranges for spending and impressions. Google’s Ads Transparency Center also provides searchable records of political advertising and information about advertisers and the audiences they reach.

These tools have limits. They help trace paid advertisements but do not provide a complete record of unpaid posts, private influencer payments or concealed arrangements between consultants and campaign intermediaries. Organic content can therefore remain outside the advertising record.

A separate example illustrates how such networks can operate. In its second-half 2025 adversarial threat report, Meta said it removed 59 Facebook accounts, 11 pages and 152 Instagram accounts from an India-origin coordinated inauthentic behaviour network targeting domestic audiences. The company reported the use of AI-generated profile photographs and biographies, apparently ordinary lifestyle posts and a separate cluster of accounts designed to artificially boost engagement.

That network was not identified as part of the Kenya or Michigan operations. It nevertheless documents techniques that can make coordinated activity appear more like genuine public participation.

What Do US Election Rules Say About Foreign Campaign Work?

The presence of an Indian consultant in a US campaign does not automatically establish a breach of election law. The legal position depends on the nature of the work, how it is paid for and whether foreign nationals participate in campaign decisions.

The US Federal Election Commission prohibits foreign nationals from making contributions or expenditures in connection with federal, state or local elections. It also bars them from directing, controlling or participating in decisions about election-related activities.

There is an important distinction for commercial services. FEC legal analysis recognises that a foreign or domestic company may provide services to a political committee in the ordinary course of business and at the usual and normal charge without that service itself constituting a contribution. This is not a blanket exemption for all outsourced campaign work: the restrictions on foreign financing and participation in campaign decision-making still apply.

The Foreign Agents Registration Act (FARA) is a separate disclosure law. It can require registration when someone acts in the United States as an agent of a foreign principal and undertakes specified political, publicity or related activities. It is not a blanket registration requirement for every overseas company providing services to a US candidate.

Platform rules also regulate the methods used to distribute political material. X prohibits coordinated inauthentic activity, including operating multiple accounts to artificially boost posts, trends or engagement. Meta and Google require identity verification and disclosures for political advertising in relevant jurisdictions. These rules do not mean that every coordinated campaign is unlawful, but they distinguish legitimate account management and advertising from deceptive manipulation.

How Far Has India's Political Campaign Industry Expanded?

The investigation describes reported work or discussions involving Indian consultancies in several countries. Besides Kenya and the US, consultants told the publication about work in Nepal and Sri Lanka. A researcher cited in the report said they had identified consultancies operating in Australia and Germany.

A proposed engagement with a presidential campaign in Suriname reportedly did not proceed. These accounts do not carry the same evidentiary weight as the Kenya and Michigan examples: some are based on interviews with consultants or researchers rather than publicly available contracts, payment records or confirmed campaign links.

The investigation also quotes consultants alleging that payments can pass through overseas companies and informal hawala networks. These claims point to the difficulty of tracing financial relationships, but the report does not provide public transaction records establishing the route of payments for the specific Kenya or Michigan activity.

The available evidence does not establish the size of the industry, the number of Indian firms involved, a common command structure or the total money spent. Nor does it show that these operations have determined an election result. The seven million views reported in the Michigan case, for instance, do not represent seven million unique voters or demonstrate that viewers changed their political preferences.

What the cases reveal is a market in which political content production and online distribution can be outsourced across borders, with intermediaries and apparently independent accounts obscuring who is behind the message. The central question is whether voters can identify who is paying for political content, who is coordinating its circulation and whose interests it serves.