The program must help recognize banished fraudsters who create new accounts to broadcast a shower of offensive, racist or sexist removal.
Le Monde with AFP
Always confronted with the problem of online harassment, and sometimes accused of inaction in the face of the violence of some users, the Twitch video game platform announced, Tuesday, November 30, the introduction of a system to detect Malicious Internet users who create new anonymous accounts to return to chains that banished them.
The leader in the dissemination of live video games for months for months to stem a wave of racist and homophobic harassment, which consists of hate raids – “hate raids” – against certain creators, including non-white persons or LGBTQ.
Harrowers come on the chat windows of their victims and flood them with shocking insults or images – for example: swasting crosses when the player is Jewish. If the creator banishes them, some come back anyway, creating a new account.
The new tool, baptized suspicious user detection, “is there to help you identify these users according to certain signals (…) so that you can take action,” said Twitch in a statement.
Thirty million visitors a day
This program, based on so-called Learning machine technologies (automated software learning, a category of artificial intelligence), will distinguish between “probable” or “possible” fraudsters. In the first case, their messages will not appear in public, only the player and his moderators will see them. Charge for them to decide then to monitor or ban them. “No Machine Learning system is 100% reliable,” says Twitch, “That’s why [the tool] does not automatically exclude all fraudsome potentials”.
The platform belongs to the Amazon technology giant, which largely dominates the global cloud industry. She claims to accommodate more than 30 million visitors a day.
In August last August, players mobilized to call the company to react to the raids. Twitch has launched new tools and has also filed a complaint against two users who, according to her, manage from Europe multiple accounts on the platform under different identities and would be able to “generate thousands of bots [automated computer programs In a few minutes “to harass their victims.