Can Twitter Manage Abuse Of Its New Voice Tweets?

Probably not right now, but when has that ever stopped social media? JL

Cat Zakrzewski reports in the Washington Post:

Twitter just rolled out voice tweets, which allow users to add up to 140 seconds of audio. But tech experts warn that the audio feature could be a new vector for harmful content that's even harder to police than text. The new feature could raise new problems as the upcoming 2020 election, civil unrest and coronavirus pandemic are raising the stakes in Big Tech's fight against disinformation. “Moderating audio takes more resources because it takes more manual analysis and currently has less options for applying automated tools.”

Twitter just rolled out voice tweets, which allow users to add up to 140 seconds of audio. But hold your jokes about how this is just the new voicemail: Tech experts warn that the audio feature could be a new vector for harmful content that's even harder to police than text. 

The new feature could raise new problems as the upcoming 2020 election, civil unrest and coronavirus pandemic are raising the stakes in Big Tech's fight against abuse and disinformation on its platforms. 

“Unfortunately, if the company gives users a new way to express themselves, some of those users are likely to take advantage of the situation to spew the sort of harmful content that has long been associated with Twitter,” said Paul Barrett, deputy director of the NYU Center for Business and Human Rights. 

Twitter's handling of disinformation, violence and hate speech is already under the microscope. 

Voice tweets will add “an additional layer of complexity” for Twitter, which has faced criticism for not doing enough to police falsehoods and other harmful content in traditional tweets, said John Redgrave, the chief executive of the start-up Sentropy, which makes content moderation tools. 

“Any new medium for distribution creates additional moderation challenges,” Redgrave said. 

Twitter is only making it available to a limited number of people initially as it works out these issues. 

“We’re working to incorporate additional monitoring systems ahead of bringing this to everyone,” said Twitter spokeswoman Katie Rosborough. “We’ll review any reported voice tweets in line with our rules, and take action, including labeling as needed.”

But Twitter has been criticized for being inconsistent about how it enforces its rules for regular tweets, and audio could be an especially challenging decision if it's coming from public figures. 

“If it's someone like Trump's account, it becomes harder to ascertain which of that content is legitimate,” said Ashkan Soltani, former chief technologist of the Federal Trade Commission. 

Twitter particularly has taken a hard line on manipulated media, at times appending labels to videos in tweets that have been doctored in certain circumstances. The new audio feature could open it up to greater audio manipulation at a time when sound-editing technology is rapidly advancing.

There are technical reasons why audio has also traditionally been more challenging for tech companies to monitor than text. 

“Moderating audio takes more resources because it takes more manual analysis and currently has less options for applying automated tools,” said Graham Brookie, director and managing editor of the Digital Forensic Research Lab at the Atlantic Council. 

To analyze audio, human moderators may have to listen to the file in full, which at 140 seconds in length could be much more time consuming than scanning a 140-character tweet. They also may rely on transcription services to convert the audio to text, which could then be reviewed by humans or algorithms, Redgrave told me. 

Given the current available technology, there is “no doubt” this will make Twitter’s challenge to moderate content more difficult, Dipayan Ghosh, a former Facebook employee who recently wrote the book “Terms of Disservice,” told me. 

But this won’t be an entirely new challenge for Twitter. 

Brookie points out that Twitter already allows video on its service, so it has dealt with similar issues with that content. And other companies that allow shared audio on their platforms are dealing with this, too. 

It also remains to be seen how widely the new voice tweet feature is used. Already, it’s become something of a punchline on Twitter.