Defamation and the Internet: Name Calling in Cyberspace

"Law and the Internet - Regulating Cyberspace" Defamation and the Internet: Name Calling in Cyberspace by Lilian Edwards 1 In recent years, defamation or libel on the Internet has become one of the hot topics of Internet law. Many gallons of both real and virtual ink have been spilled in computer and legal journals, as well as on-line electronic fora, as the impact of "terrestrial" defamation law on both suppliers and consumers of Internet services has been debated 2 This article will not attempt a comprehensive treatment of the area 3 but will focus on two crucial points: why users of the Internet are more likely than ordinary citizens to be found publishing comments which are actionable as defamatory, and what problems (or, looking at it from the other side of the fence, opportunities) arise if those who are the victims of defamatory comments on the net attempt to seek compensation by taking legal action. Two preliminary points are worth emphasising at the start. First, any lawyer looking at the problem of Internet libel is immediately struck by the fact that it is has an inherently transnational nature. Because of the international connectivity of the Internet, its speedy transmission of huge amounts of data simultaneously to multiple destinations, and general lack of respect for national borders, it is extremely easy for an individual to make a defamatory comment via a computer situated in (say) Scotland attached to the Internet, which can then be read by thousands if not millions of people similarly equipped in multiple other national jurisdictions - where (as discussed below) the law of, and defences to, defamation may be very different than those found in the Scottish legal system. In pre-Internet days, such transnational publication would have, for economic reasons, been almost exclusively the preserve of a traditional publisher, such as a newspaper, TV station or book publishing house, who would be likely to have both the resources and the foresight to take legal advice, and to have a system of prior checking in place, to avoid incurring exactly such legal liability. Very few of the individuals now setting up home page Web sites, contributing to newsgroups, sending email or taking part in Internet Relay Chat (IRC) will have such defensive strategies or knowledge of what speech might be legally actionable. Where defamatory statements cross national boundaries, inevitably problems of international private law are invoked, with difficult questions raised such as what country (or countries) will have jurisdiction to hear any action for damages raised, what country's law should govern the action (the choice of law question) and if a decree is obtained, how can it be enforced if the defender lives outwith the jurisdiction of the court (as will frequently be the case)? Those libelled on the Internet may find then that their case is not the simplest to pursue. By way of comfort, however, Internet libel defenders may also be dismayed to find that they can be sued in the courts of multiple countries to which they have little or no connection, and where the law applied is foreign to them in the extreme. Hardened libel lawyers will say there is nothing very new here, which is, formally, true - but the problems of traditional publishing and defamation are so multiplied when applied to a forum as large, as accessible, as cheap and as transnational as the Internet, that it is not hard to see why there is a perception that the law of libel has been transformed by its application to the new electronic highway. Secondly, Internet users cannot be regarded as a homogenous group. In particular, it is important to separate out the potential liability of those who give individuals and corporations access to read, and write to, the Internet : not just the so-called Internet Service Providers (ISPs) such as CompuServe, Demon, Pipex, America On-Line et al, who typically provide access to the Internet on monthly subscription or at an hourly rate, but also non-commercial hosts such as universities, who give Internet access for free to many students and staff, and corporate hosts, who have an Internet link (whether dedicated, or via an ISP) and allow their employees, and perhaps their clients, access to the Internet via their facilities. The particular problems of Internet Service Providers (and equivalent hosts) in this area are considered below. Sites of defamation on the Internet In considering why the Internet is a defamation prone zone, there are at least four distinct sites where defamation may occur on the Internet that can usefully be separated out, as to some extent they raise distinct problems. (i) One to one email messages As anyone who has used email will know, it is remarkably quick and easy to use. Comments can be typed in haste and sent at the press of a button. Compared to conventional written correspondence, where there is typically time to draft the statement, print or type it out, re-read, re-draft, and then think before signing, putting the message in an envelope, attaching a stamp and putting in the post, transmission of email is virtually instantaneous and usually, once sent, is irrevocable. As a result, email correspondence is often in substance more like spoken conversation than written interaction for habitual users - hasty, ungrammatical and rash - and tends to lead parties to say things they would not only not normally commit to writing, let alone widely published writing, but would in fact often also not say in face to face interaction with the other party. Psychologically, electronic interaction combines a sort of deceptive distance - one is after all sitting safe behind a terminal in one's own office when writing - with a kind of equally deceptive intimacy. Studies and anecdotal evidence show that there is a lack of body language, eye contact or spoken cues, as there would be in conversation or on the phone, to prevent the making of inappropriate statements4 . All this means that those sending email are dangerously prone to making remarks that turn out to be legally actionable. To add insult to injury, it is very easy to repeat or forward the defamatory comments of others via email, and in the libel law of many countries, a re-publisher is just as liable as the original publisher (bar the possibility of innocent dissemination defences, discussed below)5. For example, party A receives an email concerning the foul practices of a competitor and forwards it with a few keystrokes to parties C and D who later send it on to E and F6. Only later is it discovered that the message is not true; subsequently the competitor discovers the re-publication and sues party A rather than or as well as the original author who may be (say) without funds. In this way, actionable email statements can be re-published far and wide with the speed of transmission of any other computer virus. Sending an email containing defamatory statements from person A to person B will in some legal systems not be regarded as "publication" for the purposes of libel law, since there is no communication to the public but only to the specified recipient. This is true, for example, of English law7, but not apparently, of Scots law8. However, as is true with Internet publication generally, emails can be, and often are, sent across national boundaries eg from Scotland to England, or to France or the US. As already mentioned, this may mean that the law governing any potential action may not be that of the defender's residence or domicile. Thus the risk will not go away just because the email sender (or their ISP) are resident in England. (ii) Mailing lists The format of an electronic mailing list is that various parties subscribe by email to the list, which is administered by some central host. The subject of discussion of the list may be anything from Internet law to real ale to homosexual fantasies. Usually the list is set up so that, by default, any email message sent by any one subscriber to the list, is "bounced" or "exploded out" to every other subscriber (many of whom will, as the parlance goes, "lurk" and never be known to exist to the person commenting). Mailing lists combine all the general problems of email discussed above, with some extra difficulties of their own. It is very easy for the slightly careless or inexperienced user of such a list to think they are replying only to the maker of a particular comment - but actually send their reply to every member of the list. The embarrassment factor can be considerable, particularly where the members of the list form a small professional community within which the professional reputation of the person defamed can be severely damaged. It is not a coincidence that one of the very few cases across the globe on Internet libel not settled out of court, Rindos v Hardwick9, revolved around comments made on a mailing list for academic anthropologists in which comments were made implying that Rindos, the Australian plaintiff, had been denied tenure because he was not a properly ethical researcher and was academically incompetent. (iii) Newsgroups, the USENET and discussion fora. Newsgroups are discussion fora which are made up of comments from their subscribers, sorted by subject matter. All it takes to subscribe and post comments to a newsgroup is rudimentary software, obtainable for free as shareware, and an Internet connection. Collectively, the newsgroups available to Internet users are sometimes known as the "Usenet".10 There are something like 14,000 Usenet newsgroups subscribed to en masse by millions of subscribers, located in every country where there is Internet access. As a result, any comment posted to a Usenet newsgroup is virtually guaranteed to be published, and read, within days if not hours, in many hundreds of national jurisdictions. As can be imagined, the volume of material published in these fora is enormous - one estimate is that around 4 million articles are available at any particular time. Newsgroups are even more problematic from the defamation point of view than the rest of the Internet because of what may be described as traditional "Internet culture". Until very recently - roughly, the early Nineties - the Internet was largely the domain of technophiles, students, academics and workers in the computer industry, principally in the US. These users largely accessed the Internet for free and used it for non-commercial purposes. There was a strong collective sentiment towards anarchy, libertarianism and free speech rights - and a strong corresponding dislike of corporate, governmental or legal authority or control. In this culture, full, frank and unfettered discussion known as "flaming", which was often indistinguishable from rudeness and abuse, was not only tolerated but by and large encouraged. The usual remedy for being flamed was not to post a writ for libel, but extra-legal self help - in other words, flame back. It was and is not uncommon for newsgroups to degenerate into "flame wars" - torrents of abusive comments which destroy all sensible discussion in the group. This was all very well, perhaps, when most Internet users shared a similar cultural background. But in recent years the Internet has ceased to be the domain of "netizens" and become extensively used by individuals and families, including children, who pay for Internet access and expect it to respect the same standards of decency and courtesy as other media. Even more importantly, corporate use has expanded enormously, as firms who see the Internet as a domain for commercial expansion establish their own connections and Web sites. For these users, flaming and abuse are not acceptable, not are self-help remedies, and preservation of corporate reputation is paramount. Corporate culture now seems to have firmly encountered the Internet as in July 1997, the first corporate email libel case to be publicly settled in the UK received extensive publicity. This case was brought by Western Provident Association who sued Norwich Union Healthcare for spreading untrue rumours on its internal email system about Western's financial stability11. A settlement was reached under which Western Provident paid out the not insubstantial sum of £450,00012. (iv) The World Wide Web The Web is now so large, and increasing in size so fast that it is impossible even to pin down estimates of its size. In September 1996, there were 30 million Web pages, located on 275,000 servers, indexed by the Alta Vista search engine. At around the same date, it was estimated that the Web doubled in size every 45 days13. Like newsgroups, Web sites can be accessed and read in multiple jurisdictions, and they therefore share many of the problems of transnational publication discussed above. But perhaps the major unique problem with the Web is how far it allows any individual to mimic traditional publishing at very low cost. "Home pages" can be set up which do a good job of looking like electronic journals or glossy magazines and which can be extremely attractive, with good design and graphic content. However many of the parties setting up Web sites - often fans of popular music or TV programmes, students, pressure groups, or amateur associations - are not already hard copy or traditional publishers, have no knowledge of the law of defamation or libel, and may well find themselves publishing defamatory statements without fully appreciating their potential liability14. There has already been one at least one case in the UK where proceedings have been initiated against a Web publisher for libel. In February 1996, the Poetry Society was sued for publishing a Web page in which a vanity publishing company was accused of "preying on poets who could not otherwise get their poems published". The matter appears to have been subsequently settled out of court. Interestingly, although the Poetry Society's web site at the time was itself physically hosted by the BBC server, there seems to have been no attempt made to involve the BBC as co-defenders, possibly because the aim was removal of the offending statement rather than financial compensation. Problems and opportunities for Internet libel pursuers and defenders Jurisdiction, choice of law and enforcement As mentioned above, one of the major features of Internet libel or defamation is that it will often have been transmitted across national boundaries. In such cases, it will be necessary for a plaintiff or pursuer to work out where he or she may, and perhaps may most advantageously, raise any action. Once jurisdiction is established, there is then the question of establishing choice of law. There are self-evidently crucial differences between national laws of defamation which may favour either the pursuer or defender. For example, if we take a random scenario : An individual resident and domiciled in Scotland posts a defamatory comment about a person also resident and domiciled in Scotland, but having a national reputation throughout the UK, to a Usenet newsgroup. The group is read by subscribers in many countries, including England. The defamed party wishes to sue. The obvious court in which to sue is the Court of Session in Scotland. But under the Civil Jurisdiction and Judgements Act 1982, Schedule 8 (which applies in cases between two Scottish domiciliaries) there can be jurisdiction either in the court of the defender's domicile - Scotland - or in the place where the delict is committed. Where is a delict such as defamation committed? There are two obvious interpretations - firstly, the place where the remark was originally made (the "source" of the delict); and secondly the place where the remark is "published " ie, where it is made public and has an impact on the reputation of the person defamed (the "target" of the delict). Case-law from the European Court of Justice interpreting the Brussels Convention - notably the recent referral to the ECJ from the House of Lords in the case of Shevill v Presse Alliance S.A.15- seems clearly to establish that either interpretation is a valid alternative for the purposes of fixing jurisdiction. Thus in our scenario, notwithstanding the fact that both the pursuer and defender are Scots, there is jurisdiction in both Scotland and England. Where there is both publication, and a reputation to be affected in England, the pursuer may well wish to think about suing in England, where the damages award will almost certainly be higher than in Scotland. This is legitimate forum shopping, but one important caveat must be made; another matter clarified in Shevill is that if the action is raised in England on this kind of basis, damages can only be sought in respect of damage caused to the reputation in that jurisdiction. To sue for damage caused by the defamatory statement in every jurisdiction where it was published - which could be every country where the newsgroup was read in the case of a global celebrity with a matching reputation - the action must be raised in the courts of the domicile of the defender (in this example, Scotland). It should also be noted that forum non conveniens is still a possible plea in actions involving intra-UK jurisdiction only16, although not actions between parties from different states party to the Brussels Convention17. The logical next question in this scenario is what law will govern the action. Actions for defamation are still subject to the common law requirement of "double actionability", ie, the requirement that there must be a successful cause of action under both the lex loci delicti (the law of the place of the delict) and the lex fori (the law of the forum) before the action can be succeed18. In the example chosen, both the lex fori and the lex loci delicti are English law - so double actionability is not a problem. (This is on the assumption - as seems likely but is not wholly clear - that for the purposes of choice of law, the place of the delict is also the place where damage is caused to the reputation of the victim, ie, the "target" jurisdiction19.) But the rule of double actionability can have invidious effects for the pursuer or plaintiff where two legal systems are involved, and the law differs between them. Let us vary our scenario a little: The person defamed is a public figure, eg, a media celebrity, originally an American national, but who has established his principal home in Scotland. Both pursuer and defender are resident and domiciled in Scotland. The defamatory comment, as before, is published in a Usenet newsgroup readable in many countries including Scotland, England and the United States. The principal harm done to the pursuer's reputation is in the United States. Will the action by the celebrity succeed if raised in Scotland? There is jurisdiction to sue in the place of the defender's domicile - Scotland -for the whole damage caused to the pursuer's reputation in all countries. To successfully sue for damages in respect of the damage to the reputation in the US, there must however be a successful cause of action under both Scots and US law.20. In the US, it is effectively a successful defence to an action for libel that the pursuer or plaintiff is a "public figure."21 In such cases according to US law, the burden is put on the pursuer to show by clear and convincing evidence that the defender made the comments with actual malice. In Scots law, by contrast, such malice is presumed. It is quite possible then that although the action would succeed under Scots law, the pursuer may fail as a result of the double delict rule - an example of US law controlling the result of an action between two Scots domiciliaries. The only possible line of attack for the pursuer in this example lies in the approach taken in the cases of Boys v Chaplin22 and Red Sea Insurance Co Ltd v Bouyges S.A & Others23 in which the House of Lords and the Court of Appeal, respectively, chose to approve the possibility that in appropriate circumstances the double actionability rule might be displaced in favour of a "proper law" approach. In a case of the kind above, there might conceivably be a conclusion that the "centre of gravity" of the action was in Scotland and that Scots law should be the proper law. Finally it is important to remember that winning the action is only half the battle. Where the defender in an Internet libel case lives abroad, the judgement will still need to be recognised and enforced by the courts of the defender's residence (unless he is foolish enough to leave major assets in the pursuer's country of residence). Many countries may choose not to so recognise, either because they have no clear mechanisms in place for recognition of foreign decrees, or because the legal basis of the judgement runs against principles of their own legal system, eg, an over-riding constitutional preference for freedom of expression. Such problems have arisen even in respect of judgements for libel obtained in the English courts where enforcement was then sought against a U.S. defender. 24 Liability of Internet Service Providers The key role of ISPs such as CompuServe, Demon at al is to provide access to the Internet for their subscribers. This access includes allowing subscribers both to read and write to Usenet newsgroups; and to surf the Web. ISPs also sometimes host "local" discussion fora - newsgroups accessible only by their own paid up subscribers and not therefore part of the general Usenet - and almost invariably agree to act as physical hosts to Web pages set up by their subscribers (generally to a maximum storage of a few megabytes). In all these cases, the ISP runs the risk of being regarded as the publisher of libellous remarks, originated by another person, but published by them in one of these forums. As noted above, it is clear in principle that in both Scotland and England25, any repetition or re-publication of a defamatory statement is in itself actionable. Action is possible against all intervening persons who are responsible for repeating, publishing or otherwise circulating the defamation. The person defamed may thus choose whether to sue the original defamer, or the repeat publisher, or both - and in many cases, will be best advised to sue the party with the deepest pockets, usually the ISP, rather than the original author. But as a practical issue, far too much material passes through Usenet newsgroups alone at any one time for an ISP to physically scrutinise it all in advance of publication, and it is generally impossible to exclude any particular message in a newsgroup, only the whole newsgroup. ISPs thus have almost no control over much of the material they are "publishing". Software does exist to search for and block access to material of an offensive or pornographic nature on the Internet26, but it is of little use in relation to defamation, where there are no specific words or images which can be predicted as attracting legal risk. As we have seen, someone's reputation may be savaged as easily in an amateur poetry forum as a newsgroup on bestiality or sexual fantasies. This adds up to a liability time-bomb for ISPs, which could seriously affect their ability to operate commercially, unless defences of some kind are made available to them. In both the US, the UK, and elsewhere, ISPs have tried to claim that they should be exempted from liability on the basis of concepts of innocent dissemination - essentially claiming that have no effective control over the material they re-distribute, and thus should not be held legally liable in respect of it as publishers. To some extent this argument rest on whether ISPs are seen as more akin to conventional hard copy publishers, or TV and radio broadcasters - who have control over what they publish, and a corresponding duty to check that the material they publish is not defamatory - or whether they should be seen as more like "common carriers" such as the phone company - who are seen as "mere passive conduits" for information, with no effective control over it, and who are thus usually not held liable for whatever material they carry. Somewhere between the two a third analogy can be drawn, to news-stands or bookstores - persons who are responsible for distributing large quantities of potentially defamatory material and have some chance to examine it, but who cannot reasonably be expected to check it all in detail if they are to stay in business27. Two widely discussed US cases28 have failed to settle in detail the issue of whether ISPs should have the benefit of an innocent dissemination defence.29 In Cubby v CompuServe30, CompuServe were sued in respect of a message appearing in a local forum hosted by them, called "Rumorville USA". CompuServe had employed a third party specifically to edit and control the content of this forum. The third party posted the information on the Internet once it was edited, with no intervening opportunity for CompuServe to review the material prior to publication. CompuServe argued that they were merely a distributor of the information, not a publisher, and should therefore not be held liable. The New York District Court agreed, holding that CompuServe was here acting in a way akin to a news-stand, book store or public library, and that to hold it to a higher standard of liability than these distributors, would place undue restrictions on the free flow of electronic information. But in Stratton Oakmont Inc v Prodigy Services31, the decision went the opposite way. On similar facts, Prodigy was sued in respect of comments posted to a local discussion forum it hosted. Again , Prodigy had employed persons known as "board leaders" to monitor and edit the content of the forum and had empowered these board leaders to remove material, although only after it was posted. The crucial difference from the CompuServe case (such as there was) was that Prodigy had explicitly marketed itself as "a family oriented computer network", which as part of its "value added" services, would control and prevent the publication of inappropriate messages. This seems to have been enough to lead the court to regard Prodigy as the publisher of the libels in question, rather than as a mere distributor, and accordingly they were held liable. The most unfortunate aspect of the Prodigy and CompuServe decisions is that the ratio that can most easily be extracted from the two contrasting results is that to avoid liability, an ISP should do as little as possible to monitor and edit the content of the messages or other material it carries. This, it can be argued, will make it seem more like a news-stand, and less like a publisher. Such a "head in the sand" approach is an extremely unhelpful message for improvement of Internet services, where as any user will know, one of the key problems for real commercial use is the huge volume of unedited, disorganised, misleading and often offensive text that has to be worked through to reach any useful information. What the Internet needs is more editorial control by ISPs, not less. More unfortunately still, these "head in sand" aspects of the Prodigy decision seem to be reinforced by the new legislation on defamation which came into force throughout the UK in September 1996, and had as one of its explicit aims the clarification of the defence of innocent dissemination for Internet providers in both England and Scotland32. Section 1(1) of the Defamation Act 1996 provides that: "In defamation proceedings a person has a defence if he shows that - (a) he was not the author, editor or publisher of the statement complained of, (b) he took reasonable care in relation to its publication, and (c) he did not know, and had no reason to believe, that what he did caused or contributed to the publication of a defamatory statement." Although this section is an improvement over the vagueness of the pre- existing common law, its phrasing still leaves much to be desired from the viewpoint of ISP liability. The defence of proving "reasonable care" provided by s 1(1)(b) is only available to persons who are not "publishers" according to s 1(1)(a). A "publisher" is defined in s 1(2) as a commercial publisher, ie, a person whose business is issuing material to the public. This would certainly seem to exclude non-commercial hosts such as universities, but to embrace commercial ISPs. Furthermore, if an ISP monitors or edits content, as both Prodigy and CompuServe did, it is also likely to be regarded as an "editor" as this is defined as including any person "having editorial or equivalent responsibility for the content of the statement or the decision to publish it." However s 1(3) goes on to state that "A person shall not be considered the author, editor or publisher of a statement if he is only involved … (c)in … operating or providing any equipment, system or service by means of which the statement is retrieved, copied, distributed or made available in electronic form;…[or] (e) as the operator of or provider of access to a communications system by means of which the statement is transmitted, or made available, by a person over whom he has no effective control." [parts omitted and emphasis added] It is clear that section 1(3)(e) was intended by Parliament to be the umbrellaunder which ISPs could shelter themselves from liability33. But this sub-section is problematic in that it seems to require, in a style rather reminiscent of the Prodigy decision, that to get the benefit of the s 1(1) defence, the ISP must only provide Internet access, and not do anything else - not, for example, exercise editorial control or spot-check content - for if they do, it would seem they will be exercising "effective control" over the maker of the defamatory statement. Yet it seems unlikely that an ISP which neither monitors nor edits can succeed in proving, as s 1(1)(b) requires, that it took "reasonable care" to prevent the publication of the defamatory statement. There is thus an inherent catch 22. One possible escape might lie in claiming that an ISP which edits content is only exercising effective control over the defamatory statement, not the person who makes the statement. Another approach might be to seek exemption from publisher/editor status under s 1(2)(c) rather than 1(2)(e), which although less apparently descriptive of an ISP does not contain any "hands off" requirement. If either of these arguments is accepted, what must an ISP do to be seen to exercise reasonable care? Section 1(5) provides that a court should have regard to the nature or circumstances of the publication, and in particular to the "extent of the responsibility of the defender for the content of the statement." In relation to a Usenet newsgroup, for example, where very large amounts of material arrive by the hour from all over the globe, and the system operator has almost no control except to censor the entire newsgroup, this would, one hopes, be very little responsibility at all. It is noteworthy that both the CompuServe and Prodigy cases involved local rather than Usenet discussion fora, where the ISPs had at least a reasonable chance of keeping an eye on the material complained of. Finally if all attempt at claiming a s1(1) defence fails, an ISP may wish to avail itself under ss 2 and 3 of an offer to make amends. If such an offer is accepted, further proceedings against the offending party are barred. ISPs are in a particularly good position to offer "a suitable correction" of the statement complained of and to publish it, as required by s 2(4)(b), far and wide, since they can at almost no cost distribute the apology to the whole of the Internet. Although ss 2 and 3 may be of practical use, it seems there are no panaceas to be found in s 1 of the 1996 Act. Even if the interpretation of s 1(1) does run favourably to ISPs, the Act will, of course, only operate to relieve an ISP of liability where the litigation in question is governed by the law of England or Scotland. If Demon, for example, is sued in France by a French resident for a statement posted in a Demon local newsgroup, then the defence in s 1(1) will only be relevant if UK law is the governing law of the cause, which is more than likely not to be the case. However it is not, of course, just the 1996 Act which lacks extraterritorial reach, but the whole of UK defamation law. In the end the simplest solution for ISPs afraid of being sued in the UK may be to physically locate their business overseas, in a jurisdiction with less exorbitant libel damages than England, and where foreign decrees for damages are not easily enforced. Solutions? Before considering what solutions there are to the problems identified above in relation to Internet libel, it is worth asking if there is really a need for anything more than legal inertia. In the last five or six years of frenetic Internet expansion, after all, there have been only a handful of Internet libel cases receiving international attention. It is submitted however that these cases are merely the tip of the iceberg. Because of the uncertainty of the law on innocent dissemination, and the scale of potential risk, it is likely that far more Internet libel cases have been settled out of court or by apology, than have ever even made it to the stage of serving a writ34. This artificial hiatus will not however last forever, especially as commercial enterprise on the Web becomes more prevalent. In the US, an artificial has been applied to Internet libel cases by the introduction of the Communications Decency Act 1996 (CDA), which provided criminal sanctions for Internet operators who published offensive material on the net, but also granted (in s 230(c)) immunities from liability to ISPs who publish objectionable material. In Zeran v America Online Inc.35, the Eastern Virginia District Court found that the existence of the CDA pre-empted the right of the court to hear an action for libel and failure to remove a offending statement brought against America Online. Imposition of common law liability on AOL would have frustrated the objective of s 230(c), which was to encourage ISPs to put in place monitoring and blocking controls so as to restrict circulation on the Internet of offensive material. Accordingly the action was struck out. Although the CDA has been partially struck down36 as in breach of the constitutional right of freedom of speech, the provision referred to in Zeran remains in force and has been upheld in subsequent libel litigation37. In the UK there are signs that Internet libel is not only becoming more prominent, but that the risk of suit is being spread even wider than the original author and the ISP or host site. In July 1997, we have seen not only the Western Provident case already discussed, but also the Jimmy Hill case. This concerned a Web site known as the "Tartan Army" which posted information about the Scotland football team, and was sponsored by the brewers Scottish Courage38. The site contained a forum where fans could directly post their views about "the beautiful game". Unfortunately one of the main topics of discussion was Jimmy Hill, the sports broadcaster, and various obscene, rude and defamatory comments and jokes were posted about him in a variety of languages. The most interesting point about the case, perhaps is that Mr Hill chose not to sue the Web site owners themselves, nor their ISP, but instead Scottish Courage, the sponsor.39 This is a worrying precedent for other sponsors and advertisers on the Web, who have next to no control over what is displayed in proximity to their name, and might seriously impede commercial exploitation of the Web. So what steps can be taken to reduce the risks to ISPs and other parties of being sued for Internet libel? As we have discussed above, national legislation such as the Defamation Act 1996 is of little use when attempting to regulate, and provide defences in respect of, transnational Internet libel. Self help solutions are possible, such as the imposition by contract of an indemnity against possible legal liability arising out of the acts of any person who subscribes to an ISP. Such "shrink wrap clauses in cyberspace"40 are however of limited utility: first, they will be subject (whatever the proper law of the contract is) to mandatory consumer protection rules such as the UK Unfair Contract Terms Act 1977 and the EC Directive on Unfair Terms in Consumer Contracts; secondly, they are likely to make informed consumers simply turn to another ISP in what is an increasingly competitive market for Internet services41; thirdly, and most importantly they will not provide relief where an ISP is sued in respect of a defamatory statement made by a non-subscriber but published by the ISP. Most legal (as opposed to "net lawyer") commentators accept that in this field, as in others such as breach of copyright, trademark infringement, obscenity and pornography on the net, single nation legislative strategies are pointless and that the way forward is by multilateral agreement leading to an international convention42. Certainly an international agreement on defences of innocent dissemination would be of use both to ISPs and to individuals, as would an agreement to harmonise or clarify the rules of international private law in relation to transnational torts. The crucial question, however, is whether there is sufficient political imperative to push such an international agreement into being. Not only do such agreements require abandonment of national sovereignty on matters of acute local interest such as definitions of obscenity, but there is increasing agreement in the computer industry and among politicians and businessmen that over-regulation of the Internet at this early stage of its commercial development may be harmful. Furthermore external regulation runs counter to the deep-rooted anti-regulatory culture of "traditional" Internet users and may prove to be unenforceable43. An alternative approach is international co-operation on voluntary or self- regulatory approaches to control of offensive material. Following the downfall of the CDA, it appears that the US government is leaning towards a regime of industry and private sector regulation in relation to harmful content, rather than state regulation which runs the risk of being embarrassingly felled as in breach of constitutional rights and freedoms44. A similar development can be perceived in Europe, in the Green Paper recently released by the European Commission on illegal and harmful content on the Internet45. As the European Commission recognises therein, defamatory material on the Internet is just one small part of a wider problem, which is how to control the spread of material on the Internet whose content is either harmful or actively illegal - for example, material (pictures as well as text) which is criminally obscene, blasphemous, liable to incite racial hatred, illegally copied or altered in breach of intellectual property rights, etc. The solution tentatively espoused in principle by the EC to this deluge of unwanted material is self-regulation by Internet content providers in the form inter alia of voluntary subjection to a ratings scheme. One such scheme is PICS (Platform for Internet Content Selection) which was launched in May 1996 by the WWW Consortium and provides a scheme of so-called "neutral labelling" rather like that used to describe films in TV magazines. The idea is that rather than imposing censure or censorship on Internet service providers from without, consumers may themselves recognise and screen out content that offends them. A Web site, for example, may label itself using PICS "tags", as containing adult content, bad language and nudity. This will of course instantly up its hit rate! But it will also allow parents to tag it as "not to be accessed at any cost" by their children. Although such ratings schemes may conceivably be a partial answer to the problem of pornography on the Internet, it has to be said they can do little to reduce the risk of - and liability for - on-line defamation - which almost by definition may occur where you least expect it. However there does seem to be a common appreciation discernible in the recent pronouncements of the Clinton administration and the EU that making ISPs liable for harmful content outwith their control is unfruitful. The Bonn Declaration of 8 July 1997 puts it this way: "…third party content hosting services should not be expected to exercise prior control on content which they have no reason to believe is illegal"46. Given the political will behind this sentiment, it would not be surprising if before long we may see an international solution at least to the problem of service provider liability for Internet libel, if not a cessation of the phenomenon itself. That awaits a sea-change in either human nature or Internet culture, neither of which seem alterable by legislative will alone.