Mr. Ronan Costello:

In addition to our support for regulation that protects the open Internet, we and our CEO have said for several years that the company's number one priority is to promote healthy public conversation because we understand that you cannot have freedom of expression and you cannot have an open public forum online if people do not feel safe expressing themselves, the two priorities go hand-in-hand. You cannot have one without the other.

On politicians' experience on the platform, we absolutely appreciate that this is a challenge, in Ireland and in every other country where our platform operates around the world. On the specific question of multi-account abuse, this goes to our focus on the behaviour of abusive accounts. If we detect that an individual or individuals are running several accounts with the sole purpose of distorting the conversation by manipulating hashtags or abusing others, that sets off behavioural signals for us that indicate it is something that should be looked at. That is why my statement mentioned that 68% of the content in the accounts that we ultimately remove from the platform are proactively surfaced by machine learning because the machine is able to detect those kinds of behavioural signals. We absolutely appreciate that.

As a result of the fact that we are cognisant of the experience of people like the Senator and those in high-profile positions who may have overwhelming notifications at times, we are experimenting with features such as safety mode, which was discussed a couple of weeks ago at a company event. By using behaviour-based analysis, we may be able to detect that an account is getting an awful lot of incoming messages and safety mode would enable the account to automatically block incoming messages from accounts it is not familiar with or that the owner does not follow and that do not follow the owner. In other words, where there is no reciprocal mutual relationship.