Marcus Hutchins, the security expert credited with helping stop the WannaCry cyberattack, has built an automated tool designed to keep social media trolls arguing with an AI instead of real people.
Social media users are now asking the man famously known as MalwareTech to release his creation, with users on TikTok and Threads hailing him a 'Troll slayer' and calling for him to share the code after seeing how effectively his bot can keep online harassers engaged.
WannaCry Hero Turns AI Against Trolls
Hutchins has designed a custom programme that pipes mean comments and replies from suspected serial harassers into a local artificial intelligence model, which then generates automated replies designed to keep the conversation going.
The idea came from Hutchins' frustration with the usual advice given to people facing online harassment: simply block the troll and move on.
@itsmarcushutchins Trolling trolls with AI
♬ original sound - Marcus Hutchins
In a clip shared on TikTok and Threads, Hutchins argued that ignoring trolls can leave them free to target other people who are more affected by harassment.
'Trolls tend to spray and prey,' he noted.
Hutchins' reasoning is that people best equipped to ignore trolls may simply leave the harassers looking for other targets who are more affected by the abuse.
His solution was to give repeat offenders something else to argue with: a bot that generates automated replies and is designed to keep them engaged.
@itsmarcushutchins Today is the 7 year anniversary of the day i completely derailed my entire life
♬ original sound - Marcus Hutchins
Whenever a mean comment appeared under one of his posts, Hutchins reviewed the user's account history. If it looked like they were serial harassers, he added them to a watchlist so his tool could automatically intercept their future replies and pass them to the AI.
The AI Bot Designed to Keep Trolls Hooked
The programme then sends the troll's message to a local large language model, guided by a prompt written by Hutchins.
'You are a calm, confident human being... You are dealing with a troll, and your job is to be as annoying as possible while still keeping them engaged,' the prompt says.
Hutchins explained that the strategy was intended to exploit the way he believes trolls react when their target remains calm and confident.
'Trolls are usually deeply insecure. So their job is to drag you down and make you feel bad about yourself,' he said.
'So the more calm and confident you present, the more annoyed they get. And the harder they are going to try and troll you.'
The objective was straightforward: keep the troll replying to the bot rather than moving on to another person.
'The goal of the bot is one that is going to guarantee a reply from the troll,' Hutchins said.
Trolls Trapped in Endless Arguments
Hutchins tested the system for three days, and the results provided examples of the bot keeping suspected trolls engaged.
One unsuspecting user reportedly spent about 20 minutes a day arguing with the AI before Hutchins revealed what had been happening. The account subsequently went private.
But the most striking example came when an active troll unknowingly argued with the bot directly on Hutchins' reveal thread.
Hutchins eventually stopped the automated replies and told the user: 'You are aware that this is just a bot, right?'
Rather than ending the exchange, the user continued arguing and even attempted to use a 2020 Wired profile about Hutchins against him.
Hutchins brushed off the attempt, pointing out that the information being cited consisted of details he had freely shared with the journalist.
The incident showed how effectively the bot had kept the troll engaged: even after Hutchins told him he was arguing with a bot, he continued the exchange.
Social Media Users Want Hutchins to Release the Bot
The experiment went viral on Threads, with Hutchins' reveal receiving 8,000 likes, while responses on TikTok and Threads were largely positive.
Some users dubbed him a 'Troll slayer', while others asked for a GitHub link. One commenter even suggested people would pay good money for a service based on the idea.
There were also multiple calls for Hutchins to share the script, including one user who wrote: 'Please tell me ur code is on github. Link it please.'
Hutchins has said he would like to scale the concept so that serial harassers could be diverted away from vulnerable people, but doing so would violate the platform's terms of service.
For now, the project remains an experiment rather than a publicly released anti-troll service.
The idea nevertheless offers a different approach to dealing with online harassment. Instead of trying to make trolls disappear, Hutchins' AI is designed to keep them engaged with a machine rather than a human target.
The difference is that, for as long as the bot keeps the troll engaged, there is a machine rather than a human on the other side.
