Information requests (government)
|Report||Total information requests||Non-emergency requests||Emergency requests||Percentage where some information produced||Accounts specified|
|2015: Jul 1 - Dec 31||2,673||2,115||558||79%||7,435|
|2015: Jan 1 - Jun 30||2,436||1,994||442||80%||6,324|
|2014: Jul 1 - Dec 31||1,622||1,402||220||80%||3,299|
|2014: Jan 1 - Jun 30||1,257||1,125||132||72%||1,918|
|2013: Jul 1 - Dec 31||833||733||100||69%||1,323|
|2013: Jan 1 - Jun 30||902||-||-||67%||1,319|
|2012: Jul 1 - Dec 31||815||-||-||69%||1,145|
|2012: Jan 1 - Jun 30||679||-||-||75%||948|
About the numbers
This data includes the number of government requests received for account information, as well as the percentage of requests we complied with in whole or in part from the United States. We’ve also included a high-level breakdown of requests by U.S. state (below).
As Twitter, Inc. is based in San Francisco, California, the majority of global government requests for account information we receive continue to come from the United States. 48% of government information requests came from the United States between July and December of 2015. During this same period, we saw an increase of 18% in the number of accounts affected by government information requests. When we receive requests for multiple accounts (usually due to large-scale investigations), we attempt to narrow the scope of the requests whenever possible.
Twitter remains committed to fighting for greater transparency in national security reporting. Our litigation, Twitter v. Lynch, continues to challenge the indefinite non-disclosure provisions of the Foreign Intelligence Surveillance Act (FISA). With regard to National Security Letters (NSLs), Twitter is seeking to have gag orders lifted as provided for in the amendments to the NSL statute, in order to be able to report actual numbers in a future report. If we included national security requests in this report, we would have to report in bands as specified in the USA Freedom Act. Twitter has chosen not to include national security requests received in this report.
For more details about global information requests, please refer to the latest report.
Information requests by U.S. state / territory
Breakdown by state / territory (2015: Jul - Dec):
- Alabama (26)
- Alaska (9)
- Arizona (43)
- Arkansas (6)
- California (313)
- Colorado (10)
- Connecticut (13)
- Delaware (7)
- Florida (89)
- Georgia (43)
- Guam (0)
- Hawaii (4)
- Idaho (4)
- Illinois (149)
- Indiana (29)
- Iowa (10)
- Kansas (4)
- Kentucky (15)
- Louisiana (10)
- Maine (2)
- Maryland (62)
- Massachusetts (57)
- Michigan (50)
- Minnesota (61)
- Mississippi (3)
- Missouri (23)
- Montana (3)
- Nebraska (7)
- Nevada (13)
- New Hampshire (5)
- New Jersey (82)
- New Mexico (5)
- New York (409)
- North Carolina (61)
- North Dakota (2)
- Ohio (50)
- Oklahoma (5)
- Oregon (3)
- Pennsylvania (60)
- Puerto Rico (13)
- Rhode Island (1)
- South Carolina (6)
- South Dakota (6)
- Tennessee (9)
- Texas (123)
- Utah (5)
- Vermont (2)
- Virginia (431)
- Washington (27)
- Washington, D.C. (278*)
- West Virginia (4)
- Wisconsin (19)
- Wyoming (1)
Information requests include both federal and state legal process. Requests are attributed to a particular state based on the location of the requesting office.
*The number of information requests from Washington, D.C. also includes requests from U.S. Federal offices abroad (e.g., U.S. Embassy offices).
Types of legal process
|Report||Subpoenas||Court orders||Search warrants||Others|
|2015: Jul 1 - Dec 31||54%||10%||16%||20%|
|2015: Jan 1 - Jun 30||55%||6%||20%||19%|
|2014: Jul 1 - Dec 31||57%||6%||23%||14%|
|2014: Jan 1 - Jun 30||53%||10%||26%||11%|
|2013: Jul 1 - Dec 31||55%||7%||26%||12%|
|2013: Jan 1 - Jun 30||56%||11%||23%||10%|
|2012: Jul 1 - Dec 31||60%||11%||19%||10%|
|2012: Jan 1 - Jun 30||-||-||-||-|
About the numbers
Most U.S. information requests come in one of three forms of legal process:
- Subpoenas are the most common form of legal process issued under the Stored Communications Act; they do not generally require a judge’s sign-off and usually seek basic subscriber information, such as the email address associated with an account and IP logs.
- Unlike subpoenas, court orders must be issued by an appropriate court and signed by a judge.
- As prescribed by the Fourth Amendment, warrants typically require the most judicial scrutiny before they are issued, including a showing of probable cause and a judge’s signature. A properly executed warrant is required for the disclosure of the contents of communications (e.g., Tweets, DMs).
- Requests from law enforcement that do not fall in any of the above categories. Examples include exigent emergency disclosure requests (which we now breakout in the table above) and other requests received for account information without valid legal process.
Certain types of court orders
Mutual legal assistance treaty requests
Mutual legal assistance treaty (MLAT) requests may authorize district courts within the United States to order Twitter to produce information for use in a proceeding in a foreign or international tribunal, including criminal investigations.
- 2015: Jul 1 - Dec 31: 8% of court orders and 1% of search warrants received have been explicitly identified as having been issued through MLAT procedures, coming from 12 different countries.
- Identified countries: Belgium, Canada, Chile, France, Ireland, Italy, Latvia, Mexico, the Netherlands, Spain, Switzerland, and Turkey.
- 21% of MLAT requests received originated from the Netherlands.
- 2015: Jan 1 - Jun 30: 13% of court orders received have been explicitly identified as having been issued through MLAT procedures, coming from 8 different countries.
- Identified countries: Argentina, Canada, Chile, Finland, Germany, Greece, India, Netherlands, and Spain.
- 2014: July 1 - Dec 31: 6% of court orders/warrants received have been explicitly identified as having been issued through MLAT procedures, coming from 8 different countries.
- Identified countries: Canada, Estonia, Lebanon, Netherlands, Peru, and Turkey; 39% of all identified MLATs originated from the Netherlands.
- 2014: Jan 1 - Jun 30: 18% of court orders received have been explicitly identified as having been issued through MLAT procedures, coming from 10 different countries.
- 2013: Jul 1 - Dec 31: 1% of court orders received have been explicitly identified as having been issued through MLAT procedures, coming from 5 different countries.
- 2013: Jan 1 - Jun 30: 1% of court orders received have been explicitly identified as having been issued through MLAT procedures, coming from 8 different countries.
Pen register / trap and trace orders
Originally developed to obtain phone numbers from telco providers, a pen register / trap and trace (PRTT) order (in the context of Twitter) provides law enforcement with legal authority to obtain IP address records from the account identified in the order, generally for 60 days.
- 2015: Jul 1 - Dec 31: 11% of court orders received were PRTT orders.
- 2015: Jan 1 - Jun 30: 11% of court orders received were PRTT orders.
- 2014: Jul 1 - Dec 31: 13% of court orders received were PRTT orders.
- 2014: Jan 1 - Jun 30: 8% of court orders received were PRTT orders.
- 2013: Jul 1 - Dec 31: < 1% of court orders received were PRTT orders.
- 2013: Jan 1 - Jun 30: 5% of court orders received were PRTT orders.
|Report||Percentage of requests under seal||Percentage where user notice provided||Percentage not under seal and no notice provided|
|2015: Jul 1 - Dec 31||46%||7%||47%|
|2015: Jan 1 - Jun 30||45%||9%||46%|
|2014: Jul 1 - Dec 31||49%||11%||40%|
|2014: Jan 1 - Jun 30||11%||6%||83%|
|2013: Jul 1 - Dec 31||28%||17%||55%|
|2013: Jan 1 - Jun 30||20%||19%||61%|
|2012: Jul 1 - Dec 31||20%||24%||56%|
|2012: Jan 1 - Jun 30||-||-||-|
About the numbers
As noted in our Guidelines for Law Enforcement, we notify affected users of requests for their account information unless we’re prohibited or the request falls into one of the exceptions to our user notice policy.
Percentage of requests under seal:
- ‘Under seal’ means that a court has issued an order legally prohibiting us from notifying affected users (or anyone else) about the request.
- If the order does not specify an end date for the period during which Twitter is legally barred from providing notice to our subscriber(s), we may request an amended order with a specified duration (e.g., 90 days).
Percentage where user notice provided:
- When not prohibited or under an exception to our notice policy, we send affected users notice of our receipt of a request for their information, including a copy of the legal process.
Percentage of requests not under seal and no notice provided for one or more of the following reasons:
- The request was withdrawn by the requester prior to any disclosure.
- No information was disclosed in response to the request.
- The request was defective (e.g., improper jurisdiction, no valid Twitter @username), thus no action was taken and no information disclosed.
- The request was an exigent emergency disclosure request; see our Guidelines for Law Enforcement for more about emergencies.
- Local law may prohibit us from providing notice.
|Report||Removal requests - Court Orders||Removal requests - Gov’t agency, police, other||Percentage where some content withheld||Accounts specified||Accounts withheld||Tweets withheld||Accounts (ToS)||Accounts (No Action)|
|2015: Jul 1 - Dec 31||3||98||0%||107||0||0||58||49|
|2015: Jan 1 - Jun 30||0||25||0%||71||0||0||-||-|
|2014: Jul 1 - Dec 31||6||26||0%||60||0||0||-||-|
|2014: Jan 1 - Jun 30||5||26||0%||42||0||0||-||-|
|2013: Jul 1 - Dec 31||2||6||0%||11||0||0||-||-|
|2013: Jan 1 - Jun 30||2||0||0%||11||0||0||-||-|
|2012: Jul 1 - Dec 31||2||2||0%||12||0||0||-||-|
|2012: Jan 1 - Jun 30||-||-||-||-||-||-||-||-|
About the numbers
This data includes government requests (and other complaints of illegal content from authorized reporters) we’ve received to remove or withhold content on Twitter from the United States. For more specific details, please refer to the latest report on removal requests.