Mr. Ronan Costello:

I thank the committee for its invitation to participate in this session. I am a senior public policy manager for Twitter in Europe. I will first provide our observations on the general scheme of the Bill and then outline some key elements of our election integrity work.

Twitter is an open and public platform. Our singular mission is to serve the public conversation. The public conversation on Twitter is never more important than during elections. We are a global service and we strive to think globally, so every year is an election year. We take what we learn from each election cycle and use it to improve our election integrity work worldwide. Policies that were developed prior to the European Parliament elections in 2019 were improved for the US Presidential election in 2020, and on it will go through 2021 and beyond. We daily build on our efforts to protect the public conversation and enforce our policies against deliberate attempts to mislead people and manipulate our platform. Partnerships with industry peers and civil society, as well as political parties and government officials, are critical to our success and it is in this context that we welcome today’s dialogue.

Twitter shares the fundamental objective of the Bill. We want to make elections more transparent, encourage accountability and promote an honest and informative civic conversation. We welcome the proposed establishment of an electoral commission. In countries where we have worked with such a commission, such as the UK, our experiences have been very positive. Having a single point of contact is helpful for everyone involved. During the European Parliament elections, we channelled our engagement through a dedicated election team. When the process concluded, the director-general of the directorate for communications in the European Parliament reached out to express appreciation of this work.

As regards the definition of "political purpose" currently provided in the Bill, there is a remote chance that the day-to-day campaigning of non-profit organisations and NGOs may be captured in the scope of the legislation. Additionally, this definition may capture apolitical voter education and engagement campaigns that the Department of Housing, Local Government and Heritage or the newly-established electoral commission would seek to run. We respectfully submit that such categories of promotion should not be subject to the requirements set out in the draft legislation. This would likely bring unforeseen and unintended consequences.

In October 2019, Twitter decided to prohibit political advertising outright. We took this step based on a belief that political reach should be earned, not bought. We cautioned that the effects of algorithm-optimised and micro-targeted political messaging present new challenges to civic discourse which are not yet fully understood.

As regards Twitter and elections generally, prior to an election the public policy team co-ordinates a period of preparation. We connect with political parties to provide an overview of our policies. We work around civic integrity, as well as sharing practical tips with them for staying safe on Twitter. We also engage with parties and the relevant electoral authorities to ensure the right support channels are in place to escalate concerns and report any perceived violations of our rules. This external outreach is complemented by the gathering of a cross-functional internal team of experts who work on elections.

Our civic integrity policy is a recognition of the responsibility to protect election conversations from interference and manipulation. Under the policy, we prohibit attempts to use our service to manipulate or disrupt civic processes, including through the distribution of false or misleading information about the procedures around voting. Core to our work around civic integrity are our efforts to strengthen Twitter against attempted manipulation, including malicious automated accounts and spam. The fight against malicious activity and abuse has been significantly augmented by our continuously improved capacity to leverage machine learning to recognise these harmful practices. Through machine learning, we can detect behaviour-based signals that help to scale our moderation efforts. Our capacity to deploy these techniques now means that more than one of every two Tweets that is removed for violating our rules has been proactively identified without requiring user reports.

We also recognise that addressing harms associated with misinformation requires innovative solutions. Content moderation in isolation is not particularly scalable. This is why we have sought to be creative in pioneering new approaches to labelling harmful content and experimenting with community moderation. Labels help us to clarify that a claim may be disputed and then we can link to more authoritative sources.

Finally, Twitter supports regional and global regulatory alignment around technology legislation. We urge consideration of existing initiatives that have provided a structure for regional standard-setting and co-operation, including the EU code of conduct on disinformation, the upcoming democracy action plan and the digital services Act. As debate around the world focuses on how to solve public policy challenges related to our online environment, the approach of Twitter to regulation and public policy issues is centred on protecting the open Internet and we welcome any dialogue with policymakers that advances this goal.