An Internet filter is software that removes certain network traffic, which can be done with or without informed consent of the end user. Restrictions can be applied at various levels: a government can attempt to apply them nationwide (see Internet censorship), or they can, for example, be applied by an Internet service provider to its clients, by an employer to its personnel, by a school to its students, by a library to its visitors, by a parent to a child's computer, or by an individual user to their own computers. The motive is often to prevent access to content which the computer's owner(s) or other authorities may consider objectionable. When imposed without the consent of the user, content control can be characterized as a form of internet censorship. Some filter software includes time control functions that empowers parents to set the amount of time that child may spend accessing the Internet or playing games or other computer activities.
Terminology The term "content control" is used on occasion by CNN, Playboy magazine, the San Francisco Chronicle, and The New York Times. However, several other terms, including "content filtering software", "web content filter", "filtering proxy servers", "secure web gateways", "censorware", "content security and control", "web filtering software", "content-censoring software", and "content-blocking software", are often used. "Nannyware" has also been used in both product marketing and by the media. Industry research company Gartner uses "secure web gateway" (SWG) to describe the market segment.
Companies that make products that selectively block Web sites do not refer to these products as censorware, and prefer terms such as "Internet filter" or "URL Filter"; in the specialized case of software specifically designed to allow parents to monitor and restrict the access of their children, "parental control software" is also used. Some products log all sites that a user accesses and rates them based on content type for reporting to an "accountability partner" of the person's choosing, and the term accountability software is used. Internet filters, parental control software, and/or accountability software may also be combined into one product.
Those critical of such software, however, use the term "censorware" freely: consider the Censorware Project, for example. The use of the term "censorware" in editorials criticizing makers of such software is widespread and covers many different varieties and applications: Xeni Jardin used the term in a 9 March 2006 editorial in The New York Times, when discussing the use of American-made filtering software to suppress content in China; in the same month a high school student used the term to discuss the deployment of such software in his school district.
In general, outside of editorial pages as described above, traditional newspapers do not use the term "censorware" in their reporting, preferring instead to use less overtly controversial terms such as "content filter", "content control", or "web filtering"; The New York Times and The Wall Street Journal both appear to follow this practice. On the other hand, Web-based newspapers such as CNET use the term in both editorial and journalistic contexts, for example "Windows Live to Get Censorware."
Types of filtering Filters can be implemented in many different ways: by software on a personal computer, via network infrastructure such as proxy servers, DNS servers, or firewalls that provide Internet access. No solution provides complete coverage, so most companies deploy a mix of technologies to achieve the proper content control in line with their policies.
Browser based filters Browser based content filtering solution is the most lightweight solution to do the content filtering, and is implemented via a third party browser extension.
E-mail filters E-mail filters act on information contained in the mail body, in the mail headers such as sender and subject, and e-mail attachments to classify, accept, or reject messages. Bayesian filters, a type of statistical filter, are commonly used. Both client and server based filters are available.
Client-side filters This type of filter is installed as software on each computer where filtering is required. This filter can typically be managed, disabled or uninstalled by anyone who has administrator-level privileges on the system. A DNS-based client-side filter would be to set up a DNS Sinkhole, such as Pi-Hole.
Content-limited (or filtered) ISPs Content-limited (or filtered) ISPs are Internet service providers that offer access to only a set portion of Internet content on an opt-in or a mandatory basis. Anyone who subscribes to this type of service is subject to restrictions. The type of filters can be used to implement government, regulatory or parental control over subscribers.
Network-based filtering This type of filter is implemented at the transport layer as a transparent proxy, or at the application layer as a web proxy. Filtering software may include data loss prevention functionality to filter outbound as well as inbound information. All users are subject to the access policy defined by the institution. The filtering can be customized, so a school district's high school library can have a different filtering profile than the district's junior high school library.
DNS-based filtering This type of filtering is implemented at the DNS layer and attempts to prevent lookups for domains that do not fit within a set of policies (either parental control or company rules). Multiple free public DNS services offer filtering options as part of their services. DNS sinkholes such as Pi-Hole can be also be used for this purpose, though client-side only.
Search-engine filters Many search engines, such as Google and Bing offer users the option of turning on a safety filter. When this safety filter is activated, it filters out the inappropriate links from all of the search results. If users know the actual URL of a website that features explicit or adult content, they have the ability to access that content without using a search engine. Some providers offer child-oriented versions of their engines that permit only child friendly websites.
Parental controls Some ISPs offer parental control options. Some offer security software which includes parental controls. Mac OS X v10.4 offers parental controls for several applications (Mail, Finder, iChat, Safari & Dictionary). Microsoft's Windows Vista operating system also includes content-control software.
Reasons for filtering
The Internet does not intrinsically provide content blocking, and therefore there is much content on the Internet that is considered unsuitable for children, given that much content is given certifications as suitable for adults only, e.g. 18-rated games and movies.
Internet service providers (ISPs) that block material containing pornography, or controversial religious, political, or news-related content en route are often utilized by parents who do not permit their children to access content not conforming to their personal beliefs. Content filtering software can, however, also be used to block malware and other content that is or contains hostile, intrusive, or annoying material including adware, spam, computer viruses, worms, trojan horses, and spyware.
Most content control software is marketed to organizations or parents. It is, however, also marketed on occasion to facilitate self-censorship, for example by people struggling with addictions to online pornography, gambling, chat rooms, etc. Self-censorship software may also be utilised by some in order to avoid viewing content they consider immoral, inappropriate, or simply distracting. A number of accountability software products are marketed as self-censorship or accountability software. These are often promoted by religious media and at religious gatherings.
Technology
Content filtering technology exists in two major forms: application gateway or packet inspection. For HTTP access the application gateway is called a web-proxy or just a proxy. Such web-proxies can inspect both the initial request and the returned web page using arbitrarily complex rules and will not return any part of the page to the requester until a decision is made. In addition they can make substitutions in whole or for any part of the returned result. Packet inspection filters do not initially interfere with the connection to the server but inspect the data in the connection as it goes past, at some point the filter may decide that the connection is to be filtered and it will then disconnect it by injecting a TCP-Reset or similar faked packet. The two techniques can be used together with the packet filter monitoring a link until it sees an HTTP connection starting to an IP address that has content that needs filtering. The packet filter then redirects the connection to the web-proxy which can perform detailed filtering on the website without having to pass through all unfiltered connections. This combination is quite popular because it can significantly reduce the cost of the system.
There are constraints to IP level packet-filtering, as it may result in rendering all web content associated with a particular IP address inaccessible. This may result in the unintentional blocking of legitimate sites that share the same IP address or domain. For instance, university websites commonly employ multiple domains under one IP address. Moreover, IP level packet-filtering can be surpassed by using a distinct IP address for certain content while still being linked to the same domain or server.
Gateway-based content control software may be more difficult to bypass than desktop software as the user does not have physical access to the filtering device. However, many of the techniques in the Bypassing filters section still work.
Content labeling Content labeling may be considered another form of content-control software. In 1994, the Internet Content Rating Association (ICRA) — now part of the Family Online Safety Institute — developed a content rating system for online content providers. Using an online questionnaire a webmaster describes the nature of their web content. A small file is generated that contains a condensed, computer readable digest of this description that can then be used by content filtering software to block or allow that site.
ICRA labels come in a variety of formats. These include the World Wide Web Consortium's Resource Description Framework (RDF) as well as Platform for Internet Content Selection (PICS) labels used by Microsoft's Internet Explorer Content Advisor.
ICRA labels are an example of self-labeling. Similarly, in 2006 the Association of Sites Advocating Child Protection (ASACP) initiated the Restricted to Adults self-labeling initiative. ASACP members were concerned that various forms of legislation being proposed in the United States were going to have the effect of forcing adult companies to label their content. The RTA label, unlike ICRA labels, does not require a webmaster to fill out a questionnaire or sign up to use. Like ICRA the RTA label is free. Both labels are recognized by a wide variety of content-control software.
The Voluntary Content Rating (VCR) system was devised by Solid Oak Software for their CYBERsitter filtering software, as an alternative to the PICS system, which some critics deemed too complex. It employs HTML metadata tags embedded within web page documents to specify the type of content contained in the document. Only two levels are specified, mature and adult, making the specification extremely simple.
By country Australia The Australian Internet Safety Advisory Body has information about "practical advice on Internet safety, parental control and filters for the protection of children, students and families" that also includes public libraries.