If you're looking to follow news and advocacy about an anticipated Vermont legislature vote this week on legalizing marijuana, a search for the latest tweets that use the combined terms "Vermont" and "marijuana" will for many Twitter users yield zero results.

Same goes for searches for tweets using the terms "pot," "weed" or "cannabis."

The latest results for "jackass" and "jerk" – words generally printed without censorship by news outlets – also yield a blank page with the message: "Nothing came up for that search, which is a little weird. Maybe check what you searched for and try again."

The omissions appear to be the product of new default censorship by Twitter, with users required to opt out of a filter to see uncensored results. It's not clear when specific search-term restrictions took effect or how decisions on restricted terms were made. 

"Top" results for restricted terms still appear, but results for the most recent posts and for photos, videos and news content tabs do not.

STEVEN NELSON FOR USN&WR

A trial-and-error U.S. News review revealed various and curious search-term restrictions.

The latest tweets about "heroin" and "dope" are off-limits, but results for "OxyContin" – a legal pharmaceutical drug that often serves as an on-ramp to heroin addiction – aren't censored.

Searches about "kratom," a legal Southeast Asian leaf product that users say can be used to treat pain, enhance energy and resist the pull of addictions, are restricted. But searches for "kava," a legal and mildly intoxicating root product from the South Pacific, are not.

Twitter described some changes in a March 1 release, saying it was working "to make Twitter safer, moving faster than ever to do so." An earlier Feb. 7 post said it was "working on 'safe search' which removes Tweets that contain potentially sensitive content."

A flurry of user reaction around March 9 focused on Twitter's flagging of accounts as potentially "sensitive," which yielded complaints about profile-flagging from apparently ordinary users. Those cases resulted from testing a potential feature, Gizmodo reported. And while the social network previously could paste warnings on individual posts, recent tweets flagged include a safe-for-work selfie and an athletic feat.

The default censorship of marijuana-related search terms, meanwhile, attracted the attention over the weekend of activist Tom Angell, who started a petition that notes results for the term "opioids" remain unfiltered.

"Censoring marijuana-related searches prevents serious people from communicating about one of the most prominent policy issues of our time," Angell's petition says. 

Indeed, eight states have laws allowing for regulated recreational pot markets and more than half allow the drug's medical use, though recreational programs in particular may be threatened by Attorney General Jeff Sessions, who has repeatedly condemned relaxed cannabis laws. Reformers concerned about a potential federal crackdown are leaning heavily on Twitter to organize an April 24 smoke-out near Congress.

Though the restrictions have ruffled feathers, they can easily be circumvented. 

Results appear for censored terms that are preceded by a hashtag and users seeking uncensored access to the latest news on "sensitive" topics can do so without much effort – by performing a search, then clicking a settings icon that allows removal of the default filter. (Instructions can be found here.)

STEVEN NELSON FOR USN&WR

Twitter spokespeople did not promptly respond to emails seeking more information, and it's unclear how widely applied the new settings are.

The social network is increasing its efforts to censor content at the same time as other social networks.

Facebook-owned Instagram, the popular photo-sharing social network, announced last week it will add its own sensitive content filter. As with Twitter's new media and profile warnings, Instagram's notice can be eliminated with one click.

Voluntary censorship can be controversial, with advocates pushing companies for and against restrictions. Google-owned YouTube last week apologized for confusion created by censorship of videos with gay, lesbian and transexual themes on computers via a restricted mode common for computers in libraries and schools.

YouTube in January deleted clipsdocumenting an apparent anti-white hate crime in Chicago, but in the past has pointed to a massive volume of video uploads to explain why videos about some crimes go undetected. 

Facebook, meanwhile, has controversially banned nipples and some state-legal pot businesses, citing community standards, but has been scrutinized over the live-streaming of alleged crimes includingrapes.

In the company's March 1 post, Twitter official Ed Ho presented new filtering changes as up for debate, writing: "We're learning a lot as we continue our work to make Twitter safer – not just from the changes we ship but also from the mistakes we make, and of course, from feedback you share."