The Trump administration is proposing to change liability law to make it harder for social media platforms to censor content that is hateful or objectionable but not specifically illegal.

A Department of Justice proposal released Wednesday is the formal follow-through on the president’s threats to “strongly regulate” social media companies over his belief that they censor conservative voices. But the proposed changes, if passed by Congress, could also make it harder for social media sites to crack down on hateful or offensive content.

Currently, internet companies have broad protection under Section 230 of the Communications Decency Act from being sued for content posted to their platforms as long as they act in good faith to restrict posts that are obscene, lewd, lascivious, filthy, excessively violent, harassing, or “otherwise objectionable.”

It’s that last clause — “otherwise objectionable” — that the Trump administration argues is too broad and can be used to stifle free speech. The DOJ would rewrite the good faith protections to remove broad discretion about objectionable content and replace it with the mandate to moderate content that is believed to be illegal or promotes violence or terrorism.

The Department of Justice released only a description rather than legislative text that spells out their plan precisely. The outline says that without broad immunity, social media sites would need to be clear and explicit in their terms of service as to what can and cannot be posted. This could make it more difficult for sites to remove content that they deem objectionable but is not clearly illegal.

Matthew Feeney, director of the Cato Institute’s Project on Emerging Technologies, warned there is a lot of content that is legal speech but that people don’t want to see on their social media.

“There’s a reason why [Facebook CEO Mark] Zuckerberg doesn’t want videos of beheadings on his site,” said Feeney. “And there’s a reason why the vast majority of people on social media want an environment where a lot of legal but awful content is prohibited, like pornography or images of people being murdered. Those sorts of things.”

Increasing liability would likely make social media sites more cautious and willing to censor content, which seems to be the opposite of what the Trump administration wants, said Mark Lumley, director of the Stanford Program in Law, Science, and Technology.

“I think this has the classic problem of content moderation on the net — the government wants you to take down all the bad content and none of the good content,” he said in an email.

“But that’s impossible, not only because content moderation is hard and the scale is so immense, but because reasonable people (to say nothing of the Trump administration) can and do disagree on what is good and what is bad.”

The administration cannot amend the Communications Decency Act by itself and its proposals would need to be adopted by Congress, where there is growing momentum on the Republican right to regulate big tech immunity. The unusual coalition ranges from otherwise staunch anti-regulation conservatives to Josh Hawley, the junior Republican senator from Missouri who has made regulating social media companies a key part of his populist pitch.

Hours before the Department of Justice proposal was released, Hawley released a bill to limit Section 230 immunity for tech companies.

A partisan bill is unlikely to pass a split Congress, but the desire to reform Section 230 immunity crosses both parties. The EARN IT Act, which would condition tech company immunity on taking action against child sexual exploitation, is being pushed by both Democratic and Republican senators.

Critics have said the bill is a way to strong-arm tech companies out of providing their users — criminals and law-abiding citizens alike — the ability to send encrypted messages that would not be accessible to federal authorities.