Democratic presidential candidate Elizabeth Warren pledged Wednesday that her campaign would not share falsehoods or promote fraudulent accounts on social media, part of a new plan to battle back disinformation and hold Facebook, Google and Twitter “responsible” for its spread.

Four years after Russian agents weaponized those services in a bid to undermine the 2016 election, Warren expressed fresh alarm that Silicon Valley’s efforts only “nibble around the edges,” failing to fully combat online threats that mislead voters or stoke conflict, including among Democrats competing for the White House in 2020.

“Anyone who seeks to challenge and defeat Donald Trump in the 2020 election must be fully prepared to take on the full array of disinformation that foreign actors and people in and around the Trump campaign will use to divide Democrats, suppress Democratic votes, and erode the standing of the Democratic nominee,” said Warren, a U.S. senator from Massachusetts. “And anyone who seeks to be the Democratic nominee must condemn the use of disinformation and pledge not to knowingly use it to benefit their own candidacy or damage others.”

One of her Democratic rivals, former vice president Joe Biden, issued a similar pledge last summer, saying he would not “fabricate, use or spread data or materials that were falsified, fabricated, doxed or stolen” and would not use doctored visuals, including deepfakes, or “bot networks to disseminate messages.”

But the party as a whole has not set guidelines. State Democratic chairs last year backed a resolution on election integrity urging the Democratic National Committee to establish a “framework which would discourage and prevent … illicit campaign tactics.” But the proposal never came up for a vote before national party leaders.

Still, the DNC has stepped up its efforts to protect campaigns and state parties from deceptive online attacks, including issuing recommendations in December on combating disinformation. The advice for campaigns included tasking a digital staffer with monitoring social media traffic and being more active in pushing back against false narratives online. The party has its own team that uses online monitoring tools to track the spread of falsehoods, providing information to individual campaigns.

Along with her personal commitment, Warren said Wednesday her presidency would push for civil and criminal penalties for those who seek to mislead people online about when, where and how to vote. She also said she would seek sanctions against countries, including Russia, that attempt to undermine U.S. elections with online propaganda.

In the meantime, Warren called on Facebook, Google, Twitter and other tech giants to take immediate steps to bolster their disinformation defenses, including notifying users who have interacted with fake accounts or the manipulated content they share.

Warren’s call to action — issued days before Iowans are set to cast the first votes of the presidential election — reflects a growing trepidation among Democrats about the digital landscape in 2020.

In October, Facebook acknowledged fresh efforts by agents tied to the Russian government to seed posts, photos and videos designed to exacerbate tensions between Democratic candidates and their supporters. Along with new foreign threats, Democrats also have struggled to contend with repeated broadsides from President Trump’s campaign, which has purchased numerous ads on Facebook that contain false claims about his Democratic foes.

Seedy Twitter accounts spread a duplicitously edited video of Biden earlier this year. Warren’s opponents similarly tried to rile black voters against her candidacy using a highly doctored photo. And backers for Sen. Bernie Sanders (I-Vt.) and his 2020 bid have been some of the most aggressive online, weaponizing features on Facebook to share viral, hostile memes about his fellow Democratic contenders.

Top Democrats publicly have blasted Facebook, Google and Twitter for failing to take aggressive action against disinformation. Each company in recent years has hired more content reviewers, and with the aid of artificial intelligence, acted more aggressively to thwart what Facebook calls “coordinated inauthentic behavior.”