Today Twitter’s decision to ban political adverts takes effect. Online political advertising is under scrutiny like never before during this General Election campaign. While offering a potentially positive medium through which to engage citizens in the General Election, concerns have mounted over the past year at the potential for online advertising to be used for sinister purposes. Over recent weeks commentators have shown how campaigners have used personal data to target advertising messages to specific groups of voters online, resulting in detailed analyses of campaigners’ targeting strategies. Facebook has faced criticism for not doing the same, but has moved to increase advertising transparency.

And this week Google became the latest company to announce changes to online political advertising, with a different approach.

Google’s plans announced yesterday will limit political campaigners’ ability to target messages at certain types of audience. Now, election-related adverts across Google-owned sites and advertising spaces, including its search engine and YouTube, will only be able to be targeted by age, gender and location.

These changes aim to prevent campaigners from targeting ads at people based on their political leanings, and from matching their own databases of prospective voters against Google's user base so that ads follow individuals around the web. The company is the first to directly take action to tackle targeting, and there are signs that Facebook may be set to announce a similar idea. Google’s policy change will begin to be implemented in the UK within the next week, marking a significant change in what it is possible for campaigners to do.

Against the backdrop of these rapidly announced changes, it’s important to stop and ask whether we are comfortable with platforms’ power to impose such change.

As Twitter, Facebook and now Google have shown, companies have considerable power to alter the terms of political engagement and debate. They can change the rules for how campaigners can communicate, and they do so in accordance with their own vision that may or may not be shaped by coherent democratic ideals.

Operating as global companies, many of these organisations want to establish uniform policies that can be applied around the world. While understandable, this approach ignores the very different electoral contexts and norms that define electoral systems. Far from having uniform standards for acceptable electoral practice, countries differ in what they are willing to permit in election campaigns. These subtleties are, however, often ignored by platform companies, meaning that each new policy announcement fails to recognise and reflect local practices and ideals.

There is a danger of a two-tier system of regulation emerging, with electoral actors subject to rules imposed both by national governments and by global technology companies. At the very least, we need to see much clearer coordination, with platforms working with national governments to develop rules, rather than simply announcing changes that will have a significant and wide-ranging impact on how elections work.

With the UK in the midst of an election, the power of these companies is particularly stark. Overnight, Google has altered the rules of engagement, changing what it is possible for election campaigners to do. This change will, as Google acknowledges, take some time to roll out, but by starting with the UK, Google will force parties and campaigners to rapidly adapt to new constraints.

While we may imagine that political campaigns are well resourced, slick operations, staffed by an army of consultants and spin doctors, in practice most parties have small teams, already working long days and delivering a huge range of tasks. Google’s latest announcement adds another consideration to campaigners’ to-do list, and it is likely that the best-resourced campaigns will be those that have the capacity to respond. This latest change therefore runs the risk of a new electoral inequality that could have a significant impact on the shape of online advertising campaigns in the UK’s current election.

And although Google has set out its plans to prevent targeting practices it feels are problematic, there is no evidence that the changes they have made will have the desired effect. At the moment we have limited evidence about what impact different policy changes will have, so this announcement marks a huge experiment. While the company undoubtedly feels pressure to act in the face of concerns about voter manipulation, the most important course of action is to establish what is happening, how extensively and with what impact. We need more research (and more transparency to enable that research) before such sweeping policy changes are made.

This latest policy announcement therefore reveals the need to urgently think more about the power that platform companies have to set the terms of democratic debate. While attempts to drive change are laudable, they should not be uncritically welcomed. There are many reasons to be cautious about ceding power to companies to decide how elections can and should be run.