Between January and June 2020, Twitter received 12,700 information requests from governments and law enforcement agencies around the world, reflecting a significant increase of 44% compared to the previous six months from July-December 2019, according to Twitter’s transparency reports released on January 11. This is a record number of government information requests that the social media company has received, in addition to 25,600 accounts specified in the requests. The company complied with 36.7% of the global requests. The number of accounts specified increased by 26% and compliance rate fell by 26%, it said.
The United States has been the top requesting country since the first Twitter Transparency Report, where it accounted for 80% of all global requests in 2012. While it still sends in the highest number of requests, it accounts for only 27% of all requests and 39% of global accounts specified. India sent the second highest number of requests, making up 21% of all requests and 25% of global accounts specified. Requests from France doubled (up by 99%) and accounts specified increased by 65%.
- Government information requests: 12,657 requests, 44% higher from Jul-Dec 2019
- Routine requests: 10,425 requests, up by 50%
- Emergency requests: 2,232 requests, up by 20%
- Compliance rate: 37%, down by 9%
- Accounts specified: 25,560 accounts, up by 26%
Emergency requests involve a danger of death or serious physical harm to a person that Twitter may have the information to prevent. Examples include a report of suicide, terrorist attack or a bomb threat.
- Between January to June 2020, emergency requests accounted for one of five global information requests. Emergency requests increased by 20%, and the number of accounts specified in them increased by 24%.
User notice: Users are notified when information on their account is requested, unless Twitter is prohibited to notify them or the request falls into one of the exceptions within the company’s user notice policy. For instance, Twitter doesn’t notify users when it could be counterproductive, such as when information is requested for accounts involved in imminent threat to life, child sexual exploitation, or terrorism. It also doesn’t inform users if the legal request comes with a non-disclosure order.
- If we are not permitted to notify the user(s) at this step in the process (e.g., because the legal request is accompanied by a non-disclosure order), we may notify the user(s) about the existence of a legal request after Twitter has withheld the reported content or disclosed information associated with the Twitter account(s).
“We were able to notify account holders in response to 250 global information requests, an increase of 28% during this reporting period,” Twitter said.
Preservation requests: In response to governments and law enforcement, Twitter also temporarily preserves account information pertaining to an investigation. The accounts are preserved for a period of 90 days, “pending issuance and service of valid legal process”. Global preservation requests increased by 75%, and accounts specified increased by 180%. The United States (58%) and India (18%) together accounted for 76% of all global preservation requests.
Global Removal Requests
Twitter received a record 42,220 legal demands, specifying a record 85,375 accounts. Some or all of the reported content in 31% of these demands were met by Twitter. This means that Twitter found only a third of the legal demands legitimate per its policies. Around 96% of the total global volume of legal demands originated from only five countries (in decreasing order): Japan, Russia, South Korea, Turkey, and India.
- Legal demands: 42,220 demands, of which 1,894 were court orders; overall 53% increase in demands
- Compliance rate: 31%, down by 10%
- Accounts specified: 85,357 accounts, down by 13% from 98,595 accounts
- Accounts withheld: 82 accounts, down by 67% from 250 accounts
- Tweets withheld: 3,069 tweets, down by 13% from 3,518 tweets
- Account TOS; 25,517, of which only 407 arose from court orders; overall increase of 1%
Legal demands from Japan increased by 59%, which were primarily related to laws regarding narcotics and psychotropics, obscenity, or money lending. The next highest volume of legal demands came from Russia, over half of which were about Russian laws prohibiting the promotion of suicide. South Korea accounted for 11% of global legal demands, its largest volume of requests to date.
Information requested on journalists
Globally, 158 accounts of verified journalists and news outlets from around the world were subject to 333 legal demands, a 22% decrease in the number of accounts since the previous reporting period. The majority of these legal demands originated from India (149) and Turkey (142).
Two tweets were withheld in India under Section 69A of the IT Act 2000. In Brazil, one tweet was withheld. Twitter did not act on the remaining demands, since they fell under its protected speech policies.
Twitter filed legal objections for court orders from Turkey that involved verified journalists or news outlets, arguing that those decisions are contrary to Turkish protections of freedom of the press. None of those objections were successful during this reporting period. – Twitter Transparency Report
Country Withheld Content
This is when a tweet or account is withheld in response to country-specific laws. Law enforcement agencies, government officials, and third-parties with valid legal requests can ask for potentially illegal content or accounts to be withheld, as per Twitter’s policies. Twitter used this to withhold content in response to legal demands from 18 countries, including India, Russia, Singapore, among others.
Non government information requests
Twitter also gives information to non-government entities such as for civil actions, divorce proceedings or requests made by criminal defendants. There was a 20% increase in such requests, and a 8% increase in accounts specified. Twitters’ compliance rate to such requests fell by 38%. Twitter said it “frequently objects” to non-government requests to disclosed account information of anonymous or pseudonymous Twitter users, particularly in the US.
Online anonymity: Twitter objected to 14 U.S. civil requests for account information that sought to unmask the identities of anonymous speakers on first amendment grounds during this reporting period. We ended up litigating three of those requests. Twitter prevailed in one case, and two are still pending. No information was produced in response to the other 11 requests.
Accounts actioned under Twitter rules
These are actions based on Twitter’s enforcement of its own rules. Twitter said it’s operations were impacted due to the COVID-19 pandemic. Since its operating centers were shut in March, human review capacity fell, and reliance on automated systems increased; a trend across Big Tech. As a result, Twitter said it prioritised acting against content that could cause severe harm, such as child sexual exploitation and terrorism, and reports of rules violations, hence resulting in backlog in other areas.
- Accounts actioned: 1,9 million, down by 16%. Of these, over 635,000 accounts were suspended for hateful conduct; 444,000 for child sexual exploitation; 398,000 accounts for abuse or harassment
- Accounts suspended: 925,744, up by 6%. Of these, 438,000 accounts were suspended for child sexual exploitation, 127,000 for hateful conduct, and 120,000 for impersonation
- Content removed: 1.9 million, down by 33%. Of these, 955,000 were for hateful conduct and another 609,000 for abuse or harassment.
There was a 48% decrease in accounts actioned for violence and a 5% increase for terrorism/extremism. Similarly, there was a 34% and 35% fall in abuse/harassment content and for hateful conduct, respectively. Accounts actioned for promoting suicide or self-harm halved.
- 15% increase in number of accounts actioned for sensitive media
- 7% fall in accounts actioned for accounts actioned for regulated goods and services
Child sexual exploitation content
There was a 68% increase in accounts actioned for child sexual exploitation policy. Around 438,809 unique accounts were suspended during this reporting period for violating Twitter policies prohibiting child sexual exploitation. 91% of those accounts were proactively identified by employing internal proprietary tools and industry hash sharing initiatives. These tools and initiatives support our efforts to surface potentially violative content for further review and, if appropriate, removal.
Twitter suspended or removed content from 4,658 accounts for violations of COVID-19 misleading information. This excludes accounts where Twitter applied a warning label or message.
Privacy: 14% increase in accounts actioned under private information policy, and 48% fall in non-consensual nudity policy. There was a 37% increase in accounts actioned for civic integrity policy violations. Enforcements increased in the lead up to the US elections in November 2020. Accounts actioned for impersonation policy violations fell by 28%.
Spam and platform manipulation
Platform manipulation refers to the unauthorized use of Twitter to mislead others and/or disrupt their experience by engaging in bulk, aggressive, or deceptive activity. This prohibited activity includes, but is not limited to, spam, malicious automation, and fake accounts.
- Spam reports: 5.4 million, up by 16%
- Anti-spam challenges: 135.6 million, up by 54%
Spam reports are user reports when they believe they been followed, mentioned or DM’ed by a suspected spam account. Twitter uses anti-spam challenges to confirm that a human is operating an account, by verifying them through reCATPCHA, or email address and phone numbers. In the period, Twitter saw more usage but also more spam attempts. Twitter increased efforts to detect spam activity targeting COVID-19 and this may contributed to increase in anti-spam challenges.
- India demanded more info from Twitter due to Covid and Delhi Riots
- Relying on automated takedowns, YouTube removed more videos in April-June than it ever has