Addressing the deepfake challenge ahead of the European elections

An abstract image that says 'Check. Rechek. Vote' The latter is circled

In just two weeks, the first voters for the 2024 European Parliament elections will head to the ballot box. For the first time, the looming threat of AI-based manipulation – especially deepfakes – hangs over the polls. And, while we have yet to see deepfakes making any meaningful impact on elections, we need to be ready to tackle them. That’s why we’ve been working on helping political parties, campaigns, news organizations, and voters to prepare by arming them with tools to increase transparency, authenticate content, and safeguard the integrity of our democratic processes.

A cornerstone of our commitment is the Tech Accord to Combat Deceptive Use of AI in 2024 Elections, signed in February at the Munich Security Conference. It represents a collective effort by 27 leading technology companies to address the challenges posed by AI-manipulated content in the electoral context. This includes developing tools like Content Credentials to detect and address online distribution of deepfakes, advancing content provenance and watermarking, and working on a common approach such as the C2PA standard, driving public awareness through education campaigns, and providing transparency. This collective effort is unprecedented and crucial given the size and the scale of the threat.

Over the past few months, Microsoft has taken a number of steps to fulfil our commitments and launched new tools and actions to stay ahead of the evolving threat against voters, candidates, political campaigns, and institutions, including in Europe. As signatory to the EU Code of Practice on Disinformation, Microsoft has also introduced specific measures and actions related to the European elections, including a rapid response mechanism, and regularly reports on its comprehensive work in the Code’s Transparency Centre.

Support for political parties, media, and civil society

Political parties do not want their campaigns and candidates hijacked by disinformation or to see voters misled. We have focused much of our efforts in Europe on supporting them.

We have organized a number of awareness sessions in Brussels and across the 27 Member States with political staffers and candidates, providing them with information on the risks of deepfakes, and giving solutions to protect themselves and react effectively. Among these are Content Integrity tools, which we’ve recently expanded to all EU campaigns and news organizations to trace the origin of media and combat deepfakes. Parties can add secure “Content Credentials” to their content – giving information about who created the content, where and when it was made, and whether it has been edited. These credentials also make it easier for media to fact check content. We have also created a public website where users can report deceptive AI-generated media appearing on our consumer services.

Political campaigns, journalists, think tanks, human rights, and non-profit organizations are on the front line facing attacks and threats from bad actors. To give them an extra layer of cybersecurity protection, we offer Microsoft AccountGuard to those organizations that support healthy democracies. Today, AccountGuard is available in 23 EU Member States and in the European Parliament.

We have also set up a “Election Communications Hub” where EU governments and election authorities can connect with Microsoft security and support teams if they run into security challenges during or leading up to the elections.

In addition to being supported and protected, political parties must also take their own responsibility to abstain from producing, using, or disseminating misleading content manipulated by AI. We applaud those parties that recently signed the Code of Conduct for the 2024 European Parliament elections, under the aegis of the European Commission and Vice-President Věra Jourová, last month.

Raising awareness and supporting voter education

Empowering the public by educating them to recognize warning signs and critically evaluate digital content is the first line of defense against deepfakes. Just this week we launched a public awareness campaign in all 27 EU Member States called “Check. Recheck. Vote.” It aims at educating EU citizens about deepfakes, the importance of critically scrutinizing voter information, and the role we all have in curbing the spread of disinformation online. The campaign also promotes trusted official EU sources for voting information, as well as other government and NGO-led public education initiatives focused on the potential impact of misinformation and disinformation on elections.

As part of the AI Elections Accord, we have also committed to supporting efforts to build societal resilience and engaging with a diverse set of global civil society organizations and academics. Together with OpenAI, we recently launched a Societal Resilience Fund. It will provide grants to promote AI education and training among voters and vulnerable communities.

Particularly crucial in these last days before the European elections is the Parliament’s Use Your Vote campaign, which encourages citizens to go to the polls. We are proud to support this campaign on Bing Search by including search results tailored to the 24 official EU languages with links to official European Parliament sources and banners dedicated to the European elections, making sure that EU citizens have the right information on when, where, and how to vote.

Our commitment to defending democracy is not new and it won’t stop after the European elections. Back in 2016, we established the Democracy Forward team in response to emerging cyber threats. By 2020, we expanded the team’s focus to include threats against democratic institutions, recognizing their impact on the information environment. In 2023, we stepped up our efforts to protect election integrity with specific commitments.

Even if we won’t see mass scale AI-based manipulation in the course of these European elections, others will follow. We all need to stay vigilant and do our part to avoid AI being misused to chip away at democratic processes. We will continue to build defences and work together with society, governments, and industry, to protect the vitality and longevity of our democracies.

Note: This blog was updated on 24 May 2024 with additional information on the EU Code of Practice on Disinformation.

Tags: , , , , ,

Nanna-Louise Linde
VP, European Government Affairs