GE2025: Hundreds of fake Facebook accounts, bots spreading anti-PAP and anti-WP sentiments; MDDI urges public to be discerning

CNA

24-04-2025

Singapore
The bots being tracked by CNA have been posting misinformation, half-truths and highly one-sided opinions without substantiation that are intended to rile up genuine users on Facebook.
SINGAPORE: Anonymous Facebook profiles with inauthentic or falsified origins, also known as bots, have been active online during the 2025 General Elections and have targeted both the ruling People's Action Party (PAP) and the Workers' Party (WP), checks by CNA have found.
To this end, the Ministry of Digital Development and Information (MDDI) said on Tuesday (Apr 22) that it is looking into these accounts.
'The government is looking into these online accounts, which have been set up recently, and appear to target different political parties,' said MDDI.
'We will continue to monitor the online space for foreign interference and will act on content that is in breach of our laws. We urge the public to be discerning consumers of information.'
Over the past week, CNA detected a flood of negative comments made by unverifiable Facebook accounts aimed at certain personalities in the WP as well as the PAP.
On April 20, PAP's blog, Petir, posted two articles rebutting claims by WP prospective candidate Harpreet Singh Nehal related to the issue of a minority race prime minister.
This was followed by news coverage from various online news outlets, including Mediacorp's 8world as well as Mothership. Their articles were shared on Facebook posts that allowed users to add their comments. Mr Singh also made a Facebook post on his rebuttal to Petir.
A total of 409 instances of suspicious posts and comments were detected on April 20, originating from 149 unique profiles.
The posts were levied both at Mr Singh as well as at the PAP, including comments that state oft-repeated anti-WP or anti-PAP phrases found on online forums and social media. They appear to be aimed at an English-speaking audience and carry a similar tone – either strongly partisan statements, repeated slogans or provocative insults.
Several also include misinformation or half-truths, including some that cast aspersions on Mr Singh's credentials as a lawyer. Most include highly inflammatory opinions that lacked substantiation. Example of an inauthentic account which commented on a news article about WP candidate Harpreet Singh.
Nearly three-quarters of these posts were made on April 20 itself, with a large bulk of the posts made after Petir first published its articles at 1pm on April 20.
This was odd as Petir's article concerned an interview that Mr Singh gave to podcast Yah Lah But that was published on April 10, but CNA did not detect significant bot activity at the time.
Upon closer inspection, 90 out of the 149 unique Facebook accounts are flagged as 'fake'. The remaining 59 were inspected manually for any cues of authenticity, and all were found to be bots.
While it is still possible that a human may create these accounts and posts individually, or create spam of a partisan nature, spamming is not a typical behaviour for genuine users in a news comment section. Example of an inauthentic account that has been posting partisan content on Facebook.
For example, in one post by Mothership on Apr 20 on Petir's articles, there were a total of 151 comments from 56 authors.
This indicates that bots posted multiple comments on a single post – one bot account seen by CNA made 16 comments in about 90 minutes.
The burst of comments by a handful of accounts is also a sign of coordinated 'bot swarming' activity.
In Mothership's Facebook post for example, around 80 per cent of the 151 comments were posted around two hours after it was first published.
At one point, multiple bot accounts were detected commenting simultaneously in the first minute that the post went live. FURTHER ANALYSIS OF BOT CAMPAIGNS
Thereafter, CNA looked at other GE-related news events that also saw a hive of activity emanate from some of these bot accounts.
The user comments section of 10 Facebook posts by three news organisations, CNA, Mothership, the Straits Times, were analysed over several days. The posts links to various articles about the retirements of four PAP politicians, Mr Louis Ng, Dr Ng Eng Hen, Dr Maliki Osman, and Mr Teo Chee Hean.
Out of a total of 3,428 comments, about 47 per cent – or 1,608 posts – were flagged as originating from suspicious or fake accounts.
Upon further inspection, a total of 581 unique fake accounts were found to be responsible for a majority of these inauthentic posts. Most of these accounts posted only one comment in a single thread, creating the appearance of broad, organic participation.
These bots mainly posted either anti-PAP or anti-WP messages in the comments section.
For example, in one CNA post about Mr Ng stepping down from Nee Soon Group Representation Constituency, one user named 'Dunkie Rookie' was able to attract substantially high engagement from other users.
An example of a bot account on Facebook that has been creating partisan content. In another example in the same post, two users 'Wefo Ong' and 'Jon Cena' drove critical messages about the PAP through the use of mocking language. The posts also received high engagement. One of the bots being tracked by CNA that has been creating partisan content on Facebook.
The content in these examples are not novel. However, the accounts that posted them are inauthentic.
CNA also found 44 bots that were active in both the posts of four retiring PAP politicians as well as the earlier posts on WP's Mr Harpreet Singh.
Two of these bot accounts that had changed their Facebook profile name in the process.
An account appearing as 'Jamal Bashir' in the earlier posts on Mr Singh later altered its name to 'Vince Varun' when commenting on the PAP retirements. Another account 'Teck Beng Tan' was shortened to 'Tan TB'.
These attempts show an effort to avoid detection by other genuine Facebook users, but CNA was able to track these inauthentic accounts despite their evasion techniques.
Apart from bot activities linked to Facebook posts, CNA also discovered several accounts on TikTok that produce anti-establishment content, including some that produce content targeting Law and Home Minister K Shanmugam.
These accounts typically employ short-form videos showing image macros with AI-generated voice overs that repeat highly partisan remarks in a bite-sized format designed to be shared.
It is not known what the origin of these accounts are, or whether the owner of these accounts are located in Singapore. TELLTALE SIGNS OF INAUTHENTICITY AND BOT BEHAVIOUR
Some telltale signs of inauthentic accounts are: They were recently created, have few connections, and their posts mainly comprise a single type of content, such as those that attack a political party and its candidates. They also lack any original posts, such as photos of someone's day, that indicate the account is a real person.
Bots, on the other hand, are multiple fake accounts managed by individuals, which also suggest a more coordinated attempt to spread a certain messaging on the platform.
They are also known to make repeated comments or replies, similar to techniques used by spammers. Sometimes, multiple fake accounts end up posting similar comments, albeit with slight variations to avoid detection.
Both types of behaviour run afoul of Facebook's community standards.
CNA has sought comment from Meta, which owns Facebook. GE2025 Facebook MDDI bots fake news