In just the past week, President Trump has retweeted a QAnon conspiracy theory alleging that Joe Biden and Barack Obama had SEAL Team 6 killed, called on supporters to form an army of poll watchers, asserted repeatedly that mailing in your ballot will result in a rigged election and tweeted four times in 11 minutes falsely tying unrest in Portland, Ore., to Biden.

Trump’s place as the first Twitter president is secure. He has used the platform to govern via tweeted policy announcements, to berate adversaries both domestic and abroad, to communicate with (and occasionally disparage) his own administration officials and to exhort his supporters to action. Trump has also effectively used Twitter to distract, deflect and drive the news narrative nearly every day since he descended the golden escalator in Trump Tower to launch his campaign in 2015.

His use of digital platforms, especially Twitter but also Facebook, has enabled the world to see an authentic image of Trump, warts and all: brash, biting, abrasive and belligerent.

Increasingly, though, he is using these powerful digital platforms not only to communicate a message, nor even to send dog whistles to loyal supporters, but also as a dangerous bullhorn for causing chaos.

With millions of votes already cast and Election Day less than three weeks away, it is time for Twitter and Facebook to take away the digital bullhorn and mute the president. Trump should be silenced on both platforms until the winner of the election is determined.

I launched and led the first political advertising team at Google in 2007 before joining Twitter to do the same in 2011. My team at Twitter met with political leaders and issue advocacy organizations to introduce them to the power of this platform. Twitter, for the first time, enabled politicians to cultivate an authentic persona, maintain a direct dialogue with supporters without the filter of news media and drive the news narrative of the day. Few political leaders embraced the medium as successfully as Trump.

But now, Twitter, Google and Facebook each have implemented escalating policies seeking to contain the damage largely caused by Trump and his supporters. It’s clear they haven’t yet solved the problem.

Twitter took action early, banning political ads from its platform in October 2019. It has subsequently labeled organic, nonpaid tweets as misinformation (for example, when the president tweets misinformation about the novel coronavirus and casts aspersions on voting by mail) and, more recently, labeled as false Trump’s tweet downplaying the risks of the virus. In a remarkable and laudable step that will hobble engagement on its platform for an extended period, Twitter announced earlier this month that it would turn off key features that drive its core metrics, starting Tuesday, to stop or slow down the spread of election misinformation. Last week, in an attempt to avoid the media mistakes made in 2016, Twitter temporarily blocked users’ ability to share a New York Post link to a suspect story about Hunter Biden.

Google restricted political entities’ ad-targeting abilities and created a Transparency Report detailing political advertising. But for the most part, the company has wisely kept its head down, though that’s easier because it no longer has much of a social media network component. Google has plenty of other problems; no need to take on those of Twitter and Facebook.

Facebook, after strenuously sticking to an unrestricted-free-speech position, relented this year, first joining Twitter in labeling posts specifically intended to suppress voter turnout and then prohibiting new political ads starting the week before Election Day. Now, under relentless pressure and dealing with a president unhinged and a country on the edge, Facebook, too, has pledged to turn off political advertising indefinitely after the election. Even without the aid of advertising, however, Trump’s tweets and posts reach millions of his followers directly, before finding millions more when they are reposted and repeated across digital, print, audio and video channels.

For years, Twitter proclaimed itself the “free speech wing of the free speech party” and has subsequently leaned on the “newsworthiness” of presidential tweets to protect their presence on the platform even when they clearly violate its terms of service.

There is merit in this argument. I am proud of my time at Twitter and the efforts the company has made to enable a democratization of information and opinion never before possible. Now everyone can have a voice, and everyone else can see it, like it, comment on it and share it.

And, of course, the president’s every word is listened to, reported and scrutinized, regardless of what channel or medium he speaks on. He’s a frequent call-in guest to his preferred Fox News and Fox Business hosts, not to mention the two hours with Rush Limbaugh he managed to carve out of a “jam-packed” recent Friday. Nobody assumes Trump will stop talking if Twitter and the other digital platforms turn him off. As Twitter chief executive Jack Dorsey put it a few years ago, don’t you prefer to know what the president is thinking and doing rather than wondering what he’s up to behind the scenes and guessing what he will do next?

My experience working at Google and Twitter and speaking regularly with members of Facebook’s team is that good, smart people at all three companies are trying hard to maintain a neutral, safe platform open to a diverse array of opinion. This is for both philosophical and practical reasons: Their respective users and customers, after all, span the political spectrum. You can’t attain true Internet scale if you appeal only to one side of the aisle.

There’s another business reason to want to keep politics alive on the sites: As many studies have shown, partisan and misleading content drives engagement, a key metric that all digital platforms strive to foster. High-profile political success also showcases the power of the platforms. Trump’s use of Twitter, like it or not, has been instructive to other politicians and to corporate brands seeking to influence target audiences.

But even at the risk of infringing on the principle of free speech and decreasing engagement for a period — which is to say, hurting their own bottom lines — Twitter and Facebook must muzzle Trump as the election nears.

The platforms have already told the president that they won’t host comments suggesting that the election is being stolen, and he continued posting them. Why would anyone expect him to listen when they tell him not to question the outcome of the election or call on supporters to contest it with violence? This is someone who literally shared fake news when he recently tweeted an article about Twitter’s temporary outage from a satirical website. It is time to proactively silence him from both platforms.

In the past few weeks, between his jaw-dropping, raving debate performance and his frenetic late-night tweeting while in treatment for his coronavirus infection, the president has been entering a dangerously desperate moment. Losing an election is a personal rejection by the people and isn’t easy for any politician to swallow. Losing an election while facing bankruptcy and countless criminal investigations is much harder. The president, if the New York Times’s tax research is to be believed, is too broke to lose.

When Trump tweeted “LIBERATE MICHIGAN” in April, he didn’t need promotional ad dollars or thousands of retweets to reach his most ardent followers. And he has continued to use aggressive rhetoric about Michigan Gov. Gretchen Whitmer (D) since then — even after the arrests of men accused of plotting to kidnap her and orchestrate a coup in the state.

A recent study by the Berkman Klein Center for Internet and Society at Harvard University found that the top drivers of voter misinformation don’t come from the dark corners of social media or from Russian saboteurs. Instead, they come predominantly from the top: Trump and leading conservative news outlets repeating his digital assertions.

What will he tweet as we get closer to Election Day? Or as the returns start coming in? Is there any doubt he will question results, claim victory unilaterally or inflame his supporters to take to the streets? The stakes are too high to wait to find out.

Twitter and Facebook have indicated that they will take down any tweets that question the outcome of the election or encourage violence, regardless of who posts them. But Trump is already doing this when he questions mail-in ballots and calls for an “army” of poll watchers. The president’s erratic and dangerous behavior calls not for a label on his tweets nor the hiding of his Facebook posts, but for him to be silenced until a clear winner is announced.

Private corporations are not required to uphold the First Amendment. But even if they were, speech is not protected if you yell “fire” in a crowded theater. Trump is telling people holding tiki torches to “stand by” to set the country aflame. Why wait for him to call on armed supporters to take to the streets?

This is not unprecedented. Some nations impose a “silence period” before elections, during which campaigns pause all television and digital advertising (outside of get-out-the-vote activities). India, a nation with a sad history of violence between people and parties, uses this cooling-off period to allow voters to contemplate their decision without the cacophony of campaign communications.

In this case, however, the cooling-off period is necessary for the one candidate whose apparent sole motivation is to sow chaos and confusion around the ongoing election. Trump has already told his most dangerous supporters to “stand by.” Twitter and Facebook shouldn’t allow him to use their platforms to give violent actors the green light.