Twitter Says 99.99% of Impressions are Now ‘Healthy Content’

In its ongoing commitment to protect free speech while preserving platform health, Twitter is set to expand its ‘Freedom of Speech Not Reach’ enforcement action. The action, initially applying to the Hateful Conduct policy, will now include policies concerning Abusive Behavior and Violent Speech.

Twitter has achieved significant results since the launch of the ‘Freedom of Speech Not Reach’ initiative, demonstrating a successful balance between freedom of expression and user safety. The company revealed that over 99.99% of Tweet impressions now come from healthy content, which adheres to their community guidelines, according to its Twitter Safety team on Wednesday.

For the tiny fraction of content that does violate Twitter’s rules, the company takes proportionate enforcement action. The most serious offences, such as illegal content or harmful actors, are dealt with by content removal or account suspension.

Twitter’s enforcement approach brings unprecedented transparency to the process of limiting the reach of Tweets identified as potential policy violations, it says. This strategy includes making such content less discoverable, helping to maintain a healthier online environment.

Our commitment to our community is simple and clear – maintain free expression and make Twitter a safe space for everyone. https://t.co/yS8A2MKoCQ

— Linda Yaccarino (@lindayaX) July 12, 2023

The Twitter community has played a key role in refining this approach, offering valuable feedback to improve the accuracy of label application. These improvements include identifying instances where reach was not appropriately restricted and enhancing the recognition of context in violation detection.

The ‘Freedom of Speech Not Reach’ initiative has so far:

• Applied labels to over 700,000 violative posts under the Hateful Conduct policy.
• Reduced the reach of violative content by 81%, compared to a healthy Tweet.
• Encouraged over one third of authors to delete their Tweet after being informed of reach restrictions.
• Received appeals from only 4 percent of labeled authors, with 90 percent of all appeals responded to within 30 minutes.

While Twitter prevents ads from appearing next to labelled content, it aims to continue enhancing platform health, transparency, and safety for all users, said the company. Twitter says it pledges to provide regular updates on its progress, signalling its commitment to continual improvement and transparency.