The client who has been defamed online usually comes to us after the platform has turned him down. He has reported the post to Instagram or to Google, filled in the form, waited, and received an answer to the effect that the content does not appear to breach the platform’s rules. He wants to know what it takes to have the post removed, and how long that will take. The answer in Ontario is more settled than most people expect. The platforms will not decide whether a post is defamatory. They act on a court order that identifies the post and declares it defamatory, and, with few exceptions, they act on nothing less.
This article sets out the steps as follows, including what the platforms do with a complaint, the demand letter, the action and the motion for judgment against a defendant who does not defend, the test the court applies to a removal order, the form of order that the platforms act on, the position of a defendant who lives outside Ontario, the problem of copies and republications, and what happens after the judgment. It draws on the firm’s work in defamation and reputation management, and on the Ontario decisions of the last fifteen (15) years in which plaintiffs obtained orders of this kind.
What the platforms do with a complaint
Google’s own evidence is the clearest statement of the position. In Jeffery v. Almusslat, [2025] O.J. No. 5900 (Sm. Cl. Ct.) (“Jeffery”), a Small Claims Court action over a Google review, the court recorded Google’s evidence that its legal policy team considers every legal removal request and “can remove content that is clear and straightforward defamation,” but that “Google has chosen to defer to courts for the challenging cases regarding potential defamation before removing those reviews,” so that “it is up to the complainant to go to court in hopes of a finding of defamation regarding a post, and then, Google will voluntarily remove the post” (Jeffery at ¶ 23). The court summarized the four (4) pathways open to the subject of a review; namely, (i) asking the poster to remove it; (ii) flagging the post for Google; (iii) a formal legal request to Google supported by a sworn statement; and (iv) suing the poster, after which, “if a finding of defamation is made by a court, Google will likely remove the post” (¶ 54).
Thorpe v. Boakye, 2022 ONSC 7176 (“Thorpe”), shows the sequence in practice. When the plaintiffs sued over a set of Google reviews, Google’s lawyers told them that “if they obtained a court order declaring the reviews defamatory, Google would consider removing the reviews voluntarily” (Thorpe at ¶ 17). Google did not remove the reviews when it was served with the statement of claim itself under the Hague Service Convention, a step that “took several months” (¶ 129). When the court made orders on default “declaring the posts to be defamatory and requiring the authors of the posts to remove them,” Google, “on receiving those Orders, voluntarily removed the posts” (¶ 143).
The other platforms answer complaints the same way. Facebook’s reply to a defamation report in Braet v. Ramputh, 2022 ONSC 6682 (“Braet”), was that “it isn’t clear to us that the content you’ve reported constitutes defamation, violates your rights or is otherwise unlawful” (Braet at ¶ 11). Counsel who approached YouTube in Asselin v. McDougall, 2013 ONSC 1716, learned that “the Defendants alone could remove the comments or restrict access to them” (¶ 20). The review sites in Mirzadegan v. Mahdizadeh, 2022 ONSC 6082 (“Mirzadegan”), were “located all over the world and several declined to remove the content absent a court order” (Mirzadegan at ¶ 5), and the Arizona host of Ripoff Report told the plaintiffs in Caplan v. Atas, 2021 ONSC 670 (“Atas”), that “it would consider complying voluntarily with a court order from an Ontario court” (Atas at ¶ 126). Removal on a demand alone does happen; Change.org took down a petition on receipt of a libel notice in Lesser v. Meta Platforms Inc., 2025 ONSC 2105 at ¶ 29. It is the exception. Whether the platform itself can be sued over a review is a separate question, which we have addressed in our article on suing Google over a defamatory review.
The platforms defer to the courts for a reason that also explains why they move quickly once a judgment arrives. A host that leaves content up after notice can be liable for it. The Court of Appeal put the rule in one sentence in Warman v. Fournier, 2015 ONCA 873 (“Warman”), at ¶ 20, stating, “Liability in that circumstance turns on whether the statements at issue have been deleted by the host after reasonable notice to delete has been given,” and in Thorpe the court declined to accept Google’s argument that it could not know whether the reviews were defamatory until a court said so, finding a genuine issue as to whether Google, “upon receiving notice of them, was under a legal obligation to remove them” (¶ 146). That exposure is the reason the platforms publish a legal process at all. It does not turn a complaint into a removal.
One more distinction is important before the steps. Removal at the source and de-indexing are two (2) different things. The author or the host deletes the post; Google removes the address of a copy from its search results. Google’s own page on removal says that it “can restrict access to content that appears in Google products and services, but that content may still exist elsewhere on the web.” A judgment is drafted to do both jobs, as explained below.
The demand letter
The first step is a written demand to the author. It identifies each post by its address and quotes the words, states that they are false, demands removal by a stated date, and says what follows if the date passes. Where the publisher is an Ontario newspaper or broadcaster, the Libel and Slander Act, R.S.O. 1990, c. L.12, requires written notice within six (6) weeks after the libel came to the plaintiff’s knowledge (s. 5(1)). That requirement applies “only to newspapers printed and published in Ontario and to broadcasts from a station in Ontario” (s. 7), and the party who says that an internet post is a broadcast has to prove it (St. Lewis v. Rancourt, 2015 ONCA 513 (“St. Lewis”) at ¶ 8). For a post by an individual on social media, the claim is governed by the ordinary two (2) year limitation period.
The demand is worth sending even when it is certain to be ignored, because the refusal is the fact from which the court later infers that the defendant will keep publishing. The Court of Appeal in Warman upheld a permanent injunction on an inference drawn from, among other things, the defendants’ refusal to remove the statements “until approximately nine months after being served with the respondent’s first notice of libel and seven months after the delivery of his statement of claim” (Warman at ¶ 21). In Duncan v. Buckles, 2020 ONSC 3219 (“Duncan”), the defendant “did not stop his campaign in the face of a libel notice, after the action was started or even after the Small Claims Court found that Ms. Duncan’s fence was lawful” (Duncan at ¶ 66), and the removal order followed. A copy of the demand to the host puts the host on notice for the purposes of Warman. The refusal also goes to costs; in Manson v. John Doe No. 1, 2013 ONSC 628, Goldstein J. awarded substantial indemnity costs because the defendant “refused to even attempt to justify or remove manifestly defamatory posts in the face of a libel notice” (¶ 33).
The evidence starts the same day. Each post is captured with its address, the account name and handle, the date, and the view, like, share and comment counts, and the captures are repeated as the counts change. Two (2) refusals show what happens otherwise. In Statuscent Inc. v. Chehade, 2023 ONSC 3977, the removal order for a website failed because the hyperlinks in the record were defunct and the only screenshot “showed folders and not the specific content that was posted on the website” (¶ 12). In Kaur v. Virk, 2022 ONSC 6697 (“Kaur”), Emery J. refused to find that a WhatsApp video had been published at all, because no one had explained how WhatsApp works, stating, “The court cannot conduct its own research outside of the evidence given on a hearing to determine facts” (Kaur at ¶ 33). The affidavit on the motion has to explain each platform, how the content is seen and shared, and what the counts mean.
The action, and the defendant who does not defend
The statement of claim pleads each publication on its own, including the words, the date, the platform, the account, the address and the meanings the words bear. A general paragraph meant to sweep in everything else does not work. In Wong v. Tang, 2025 CanLII 93642 (Sm. Cl. Ct.), the court refused to treat such a paragraph as a “catchall,” stating, “It is not the role of the Court to scrutinize this material in order to identify potentially tortious content” (¶ 5). The claim is served in Ontario or, where the author is elsewhere, without leave under rules 17.02(g) and (i) of the Rules of Civil Procedure, the basis the Court of Appeal accepted for a claim to restrain internet defamation in Barrick Gold Corp. v. Lopehandia (2004), 71 O.R. (3d) 416 (C.A.) (“Barrick”), at ¶ 72. Service abroad under the Hague Service Convention takes months, as it did on Google in Thorpe (¶ 129).
Most of the defendants in these cases never defend, and the law is built for that. A defendant noted in default is deemed to admit the allegations of fact in the claim (rule 19.02(1)(a)), is not entitled to notice of any step in the action (rule 19.02(3)), and the plaintiff moves for judgment on affidavit evidence (rule 19.05). The plaintiff is not entitled to judgment merely because the facts are deemed admitted (rule 19.06). In Thorpe, the first motion for judgment failed because the copy of the claim in the motion record “omitted the very paragraphs that contained the defendants’ defamatory words” (¶ 141), and the orders Google later acted on were made only when the complete pleading was filed (¶ 143). Posts made after the claim was issued are not deemed admitted and must be proven; in Bui v. Bui, 2025 ONSC 6902 (“Bui”), the court held that such posts “are not covered by r. 19.02(1)(a), and the defendant is not deemed to admit she made them” (Bui at ¶ 9), and then found her responsible for them on the evidence. Falsity is presumed, as Sossin J. held in Emeny v. Tomaszewski, 2019 ONSC 3298 (“Emeny”), stating, “[i]n the context of a default judgment, therefore, the communications are presumed to be false” (Emeny at ¶ 23). Damages must still be proven.
An interlocutory injunction before judgment is available where the words are “clearly defamatory,” no defence of justification or fair comment has been signalled that could succeed, and irreparable harm is shown (Bagwalla v. Ronin, 2017 ONSC 6693 at ¶ 19). A mandatory order to remove content at that stage is another matter. Pattillo J. refused one in Clancy v. Farid, 2018 ONSC 7482 (“Clancy 2018”), where the defendant denied authorship, holding that an order requiring him to remove the posts “is only appropriate as a final order following a trial of the issue” (Clancy 2018 at ¶ 32). Where the defendant has not responded at all, the motion for default judgment is the shorter road to the order that the platforms act on, and it carries no undertaking as to damages.
The test for a removal order
The test was stated by Chapnik J. in Astley v. Verdun, 2011 ONSC 3651 (“Astley”), and adopted by the Court of Appeal in St. Lewis at ¶ 13, as follows, “Permanent injunctions have consistently been ordered after findings of defamation where either: (1) there is a likelihood that the defendant will continue to publish defamatory statements despite the finding that he is liable to the plaintiff for defamation; or (2) there is a real possibility that the plaintiff will not receive any compensation, given that enforcement against the defendant of any damage award may not be possible” (Astley at ¶ 21). The two (2) branches are alternatives. The Court of Appeal added that “[a] broad ongoing injunction is an extraordinary remedy which should be used sparingly. However, where there has been a campaign of defamation and a likelihood that it will continue, there is authority for such an order” (St. Lewis at ¶ 16).
On default, the courts apply that test and grant the removal order with the injunction once a branch is met. In Post v. Hillier, 2022 ONSC 3793 (“Post”), Gomery J. found that the defendant “may very well continue to publish defamatory statements,” found “a real possibility” of non-recovery given her own statements that she had no funds, and ordered her “to take down all social media posts regarding Ms. Post, within ten days of the date of this decision” (Post at ¶ 45). In Mirzadegan, Centa J. enjoined defendants in Iran from publishing defamatory statements, declined to prohibit all comment about the plaintiffs, and ordered them “to have the posts removed” (Mirzadegan at ¶¶ 17-18). In Emeny, Sossin J. reasoned that the difficulty of service and the defendant’s decision not to respond made it likely that the plaintiff could not enforce a judgment (Emeny at ¶ 61). In Duncan, the order required removal by a stated date of “all blogposts, YouTube videos and all other internet content he has posted that refer to Ms. Duncan … including comments made by others on these online posts” (Duncan at ¶ 73(c)). In Canadian Aids Treatment Information Exchange v. Blackwell, 2025 ONSC 4678, the defendant was ordered to remove “all the online material attached to the affidavit” of the plaintiffs’ witness, a copy of which was to be appended to the order (¶ 41). Gordon J. put the rationale in one sentence in Cyr v. LeBlanc, 2022 ONSC 2555 (“Cyr”), stating, “a damage award can only address past publication and an injunction is required to prevent a future spread of defamatory statements” (Cyr at ¶ 45).
The refusals are instructive, because each turned on a gap in the evidence. In Van Leeuwen v. Calderone, 2024 ONSC 7178, the posts had been removed and, in the judge’s words, “there is no evidence before me that the plaintiffs will be unable to enforce a judgment” (¶ 24). In Flegel v. Dennis, 2026 ONSC 3532, the post had come down after the libel notice and before the action (¶ 92). In Smith v. Nagy, 2025 ONSC 4629, the defendant’s refusal had been made in good faith, and the judge stated, “I am confident that the defendant will do so” (¶ 176). In Hosseini v. Gharagozloo, 2024 ONSC 1106 (“Hosseini”), Perell J. awarded $400,000.00 but made no removal order, because the platforms hosting the videos were not defendants and “there was no evidence … as to the feasibility of ordering Mr. Gharagozloo to remove or to even attempt to have the internet postings removed” (Hosseini at ¶ 86). The record on the motion has to prove continuation, the likelihood of non-recovery, the defendant’s control of the accounts, and the content itself, with current captures.
The form of order that the platforms act on
The order is the instrument, and its drafting decides whether the platforms act on it. Six (6) features recur in the orders that have worked.
The first is a declaration, in the operative part of the judgment, that each listed publication is defamatory of the plaintiff and false, and bears the stated meanings. That is the form of order Google acted on in Thorpe (¶¶ 22 and 143). The finding of falsity has a separate purpose for the American platforms. In Atas, counsel advised that “in some American jurisdictions, the plaintiffs require an affirmative finding of the falsity of the impugned publications in order to get those publications removed” (Atas at ¶ 87); Corbett J. accepted that the finding was “an appropriate ancillary remedy” although it was unnecessary under Ontario law (¶ 89), and made it (¶ 90).
The second is a schedule to the judgment listing every publication by date, platform, account and address, with the words complained of. The order in Canadian Aids Treatment Information Exchange appended the exhibited material (¶ 41); Mathen J. in CNSX Markets Inc. v. Marks, 2026 ONSC 3952 (“CNSX”), tied the removal term to “the statements identified by the Applicant in Exhibit ‘UUU’” (CNSX at ¶ 90(a)(v)); and in Thorpe a further order was needed to add an address that had been left out of the first one (¶ 22). The warning is Inivos Ltd. v. Green, 2024 ONSC 5719 (“Inivos”), where Centa J. refused to find contempt for a failure to remove earlier posts because the order was prohibitory in form, holding that “an order that is intended to be mandatory (as opposed to prohibitory), should make that requirement explicit” (Inivos at ¶ 32), and that a proper takedown term “would have explicitly listed the posts that Mr. Marsh was to remove” (¶ 33).
The third is an explicit mandatory term with a deadline, a duty to search, and written confirmation. The wording that has survived a contempt motion is Seppi J.’s in Poulie v. Johnston, 2019 ONSC 5425 (“Poulie”), requiring the defendant to “forthwith and by no later than 14 days of this Order, remove, direct and/ or cause to be removed, any and all on-line materials … from all sites” (Poulie at ¶ 33), which LeMay J. held imposes “a positive obligation to search out the information on the internet and take all reasonable steps to ensure that it is removed” (¶ 102). The Court of Appeal in Labourers’ International Union of North America, Local 183 v. Castellano, 2020 ONCA 71 (“Castellano”), upheld terms requiring the defendant to remove and destroy the material within seven (7) days and then to provide written proof of the steps taken and written confirmation that “no such material remains publicly accessible or within his possession or control, or, to the best of his knowledge, remains in the possession or control of a third party,” adding only a carve-out for use in a pending proceeding (Castellano at ¶ 22).
The fourth is a term giving the plaintiff liberty to deliver the judgment to the platforms and to anyone hosting or republishing the content, and to request removal to give effect to it. Centa J. included it in Mirzadegan, at ¶ 19, stating, “The plaintiffs may, however, request that any entity hosting these defamatory posts remove them to give effect to this decision.” Archibald J.’s order in the Clancy litigation was “a mandatory order permitting the plaintiffs to request the removal of the defamatory postings from the internet by submitting the order to the website hosts, Internet Service Providers and search engines” (Clancy v. Farid, 2022 ONSC 947 at ¶ 35), and it was made a final order “to assist the plaintiffs in obtaining its removal both in Canada and the United States” (Clancy 2018 at ¶ 33). The term does not bind the platforms and does not need to. It puts in the judgment the invitation that their legal processes respond to.
The fifth is that the judgment does not purport to order Google, Meta, X or YouTube to do anything. They are not parties, and orders directed at non-parties without proper notice have been set aside each time the point was taken. Dubé J. set aside take-down and refraining orders against Facebook in Braet because they were granted without notice, without full and frank disclosure, on “bald, conclusory and unsubstantiated” claims, and in terms so overbroad and confusing that Facebook could not comply (Braet at ¶ 80), and held that an order of that kind “must be clear, concise, and as narrow in scope as possible to achieve its specific purpose” (¶ 75). Morgan J. set aside an injunction and Norwich order against Google in Obsidian Group Inc. v. Google LLC, 2022 ONSC 1286, because an email to the address Google uses for public inquiries about service was not service; “Google must be served in accordance with the Rules, just like any other party in civil litigation” (¶ 7). The judgment operates on the author. The platforms act on it because they have chosen to defer to the courts and because a host on notice of a finding that does nothing exposes itself to liability. The Court of Appeal made the same point in Barrick, where an injunction against a defendant in British Columbia was effective because “it will operate to prevent Yahoo from continuing to post the defamatory messages” (Barrick at ¶ 76).
The sixth is precision. The injunction “must be broad enough to be effective but no broader than reasonably necessary to effect compliance” (Castellano at ¶ 18). The Court of Appeal in Castellano set aside a blanket prohibition on any comment about the plaintiffs, because the motion judge had made no finding that the defendant “was incapable of discerning between defamatory comments and legitimate criticism” (¶ 20), and amended the terms to restrain “any defamatory” statements (¶ 21). In Miranda v. Balita Media Inc., 2026 ONCA 223 (“Miranda”), the court set aside an injunction against statements “similar” to the defamatory ones as “overly broad, not proportionate and not properly defined,” because the order did not define, and the reasons did not list, the statements enjoined (Miranda at ¶ 35); the language “invites questions about whether statements are defamatory or not” and was “incapable in practical terms of enforcement” (¶ 36). The answer is to itemize. The injunction in Warman was upheld because the trial judge confined it to the statements the jury had found defamatory (Warman at ¶ 18), and the injunction in Duncan was enforced by contempt only because the reasons had set out “exactly what she found to be defamatory and hence prohibited” (Duncan v. Buckles, 2021 ONSC 1158 at ¶ 47). Within that frame the order can reach around the obvious evasions. Dunphy J. in Zhong v. Wu, 2021 ONSC 5873, applied the injunction to anything the defendant “may publish in his own name or using any other name, username, pseudonym or account over which he has control or to which he has access in fact as well as to any statements that he knowingly encourages or causes others in future to post” (¶ 70(d)(i)); the terms upheld in Castellano extend to causing “any other person to make or publish” the statements (¶ 21); and Mathen J. in CNSX prohibited tags and hashtags in any commentary about the applicant, because “[t]ags and hashtags are an important way that social media users increase the visibility of their posts to readers and other users” (CNSX at ¶ 88).
The defendant outside Ontario
The objection that an order against a defendant in another country is futile was answered by the Court of Appeal in Barrick, and the default cases have applied the answer to defendants in British Columbia, the United States, Vietnam and Iran. Blair J.A. accepted that courts “have traditionally been reluctant to grant injunctive relief against defendants who are outside the jurisdiction” (Barrick at ¶ 73), but held that a claim to restrain the publication in Ontario of a libel originating elsewhere is a recognized exception, and that the plaintiff and the publication here supplied a sufficient connection with Ontario (¶ 74). The courts, he said, can either “throw up their collective hands in despair” or “at least protect against the impugned conduct re-occurring in their own jurisdiction,” and he adopted Kirby J.’s observation that any suggestion that internet defamation must simply be tolerated is “self-evidently unacceptable” (¶ 75). The injunction may be effective even if it can only be enforced against the defendant personally when he enters the province, for two (2) reasons; namely, that it operates on the host, and that it may be enforced where the defendant lives (¶ 76). His Honour added that “the protection and vindication of that reputation in Ontario is what gives rise to the court’s mandate in cases of this nature” (¶ 77).
The later cases follow it. In Manson v. Moffett, [2008] O.J. No. 1697 (S.C.J.), Hoilett J. treated the American defendant’s belief that a Canadian judgment could not be enforced against him as a reason for the injunction rather than against it (¶ 12), and held that “the rationale expressed in Barrick is wholly applicable” (¶ 14). The Court of Appeal in Kideckel v. Kideckel, 2025 ONCA 689, refused to stay an injunction against a defendant in California pending appeal, observing that he “can act in accordance with the injunction simply by not defaming David” (¶ 18). Mirzadegan (Iran), Bui (Vietnam) and Emeny (British Columbia) are all default judgments with removal or injunction terms against defendants outside Ontario. The reverse exercise is recognized too; in Inivos, Centa J. registered English permanent defamation injunctions in Ontario and made them enforceable here (Inivos at ¶ 7). In practice, the removal order against a defendant abroad is rarely enforced by contempt, and it does not need to be. Its value is that the platforms act on it.
Copies, republications and the anonymous reposter
A removal order reaches what the defendant controls. For the rest, the orders use a best-efforts term. Nishikawa J.’s order in Elaschuk v. Thelwell, 2020 ONSC 1407 (“Elaschuk”), required the defendant to “use best efforts to remove” her statements from the third-party websites listed in a schedule (Elaschuk at ¶ 2), a term chosen “in recognition that once posted, Ms. Thelwell did not have complete control over the statements” (¶ 62), and enforced on the contempt motion by requiring “proof of written requests” to each of those sites (¶ 71(b)). Perell J.’s term in Soliman v. Bordman, 2021 ONSC 7023, was that the defendant “remove any media he controls and attempt to have removed any media controlled by others that he has published” (¶ 212).
The republications themselves are a different problem. The author is liable for them where republication was authorized or was the natural and probable result of the original publication, and in Clancy v. Farid, 2023 ONSC 2750 (“Clancy 2023”), Ramsay J. accepted that bots and replication are a feature of the medium the defendant chose, so that “the defamatory statements will perhaps always reside on the Internet” (Clancy 2023 at ¶¶ 47-48). That goes to damages against the author. It does not produce an order against the republisher, and a contempt court will not read the republication doctrine into an order against the defendant (Inivos at ¶ 30). Each republisher who is to be bound needs a claim of his own. Where the republisher is anonymous, a Norwich order against the platform to identify him is available, as it was against Google in the Manson litigation (Manson v. John Doe No. 1, 2013 ONSC 628 at ¶ 9), but the platform must be properly served (Obsidian at ¶ 7; Braet at ¶ 52) and the order “must be clear, concise, and as narrow in scope as possible to achieve its specific purpose” (Braet at ¶ 75).
One term reaches copies without a further proceeding. In Atas, Corbett J. accepted that the court can order a defendant to remove content, found that such an order would be futile against that defendant for five (5) reasons, and adopted the plaintiffs’ alternative of “vesting title to the postings in them, with ancillary orders enabling them to take steps to have the content removed” (Atas at ¶ 228). In Clancy 2023, the judgment required the defendant to assign to the plaintiffs the copyright in the defamatory content “so that the Plaintiffs may, at their sole discretion, take any legal proceedings they consider necessary to restrain the continued republication … by third parties” (Clancy 2023 at ¶ 230(d)), while a term that he “assist” the plaintiffs with removal from search engine caches and third-party sites was refused as “too vague and likely unenforceable” (¶ 239); the Court of Appeal records that the formal judgment included the assignment (Clancy v. Farid, 2024 ONCA 568 at ¶¶ 12-13). A copyright complaint by the owner of a video or a text is something every platform processes on its own terms, without any finding of defamation, which is what makes the term useful. It needs evidence. Centa J. refused a vesting order in Mirzadegan because the record held no evidence about the ownership of the accounts or about the effect of vesting on the defendants’ legitimate activities (Mirzadegan at ¶ 19).
Later posts by the same defendant are dealt with under the judgment rather than by a new action. In Caplan v. Atas, 2021 ONSC 5390, Corbett J. held that the court was not functus officio in respect of enforcement (¶¶ 9 and 11) and made an order that “does no more than provide the moving parties with enforcement of the Judgment in respect to publications made or coming to their attention after the date on which the motions for judgment were argued” (¶ 22). A judgment can say so expressly, with a term preserving the plaintiff’s liberty to move, on notice by email, to add further publications to the schedule.
After the judgment
The judgment is served on the defendant by the methods the court has validated. It is then delivered, with its schedule, through each platform’s legal removal process, with a covering letter that points to the declaration and the list of addresses. Copies that others have posted are dealt with by a libel notice to the republisher and to the host, enclosing the judgment, and, where an assignment of copyright was granted, by a copyright complaint. Every request and every response is recorded against the schedule, because the same record is needed if the judgment ever has to be enforced.
Contempt is the court’s answer to a defendant within its reach who ignores the order. Myers J. sentenced the defendant in Paramount Fine Foods v. Johnston, 2021 ONSC 6558, to “total consecutive terms of 18 months for contempt of court” (¶ 127), and in Poulie v. Johnston, 2022 ONSC 5186, the warrant for the same defendant’s arrest was “to be specifically brought to the attention of the RCMP and the Canada Border Services Agency” (¶ 78). Against a defendant who stays abroad, contempt has no practical effect unless he enters Canada, and the plan should not depend on it.
Two (2) further terms are sometimes worth asking for. A retraction was ordered in Post, where the defendant was a public figure with a significant following, in the form of a notice on her Twitter account for sixty (60) days “acknowledging that her statements about Ms. Post were false and defamatory, and providing a link to this decision on CanLII” (Post at ¶ 46). An apology is another matter; the courts have refused to order one from a defendant who would not mean it, and Perell J. called an apology from a non-resident “a futile order” (Hosseini at ¶ 87). The judgment itself is a public document, and the plaintiff “may post this court’s judgments or links to this court’s judgments as he may be advised” (¶ 87). That is often the most durable answer to a copy that resurfaces.
The damages in these cases are substantial, and they are awarded on default. The defendant in Emeny was ordered to pay $450,000.00 (Emeny at ¶ 64); the award in Hosseini was $400,000.00 (¶ 85); the general damages in Mirzadegan were $200,000.00 with $50,000.00 in aggravated damages (¶¶ 14 and 16); the awards in Post and Duncan were $85,000.00 and $70,000.00 (Post at ¶ 2; Duncan at ¶ 73(b)). Costs follow at a higher scale where the defendant refused to remove in the face of a demand; Gordon J. awarded full indemnity costs in Cyr on the basis that plaintiffs in internet cases “are put to additional expense in these types of cases and do so to protect their reputation in the community” (Cyr at ¶ 49), and Perell J. fixed substantial indemnity costs of $180,000.00 in Hosseini (¶ 88).
Practical considerations
For the person who has just found the post, the order of work is as follows. Capture everything now, with the address, the account, the date and the counts, before the post changes or disappears. Send the demand with a deadline, to the author and to the host, and keep the response or the silence. Use the platform’s complaint process, because the refusal is itself evidence, but do not wait on it. Expect months rather than weeks between the claim and the order, longer where the author is abroad and the claim travels under the Hague Service Convention, and plan the evidence for the motion from the start, because the refusals in the cases are refusals on evidence.
Be clear about what the judgment achieves. The posts come down from the accounts the defendant controls, and the platforms act on the declaration for the copies they host. Google removes the addresses from its results, and its own page says that the content “may still exist elsewhere on the web.” Some copies will persist, as Ramsay J. accepted in Clancy 2023 (¶ 48). The judgment that declares the statements false is the answer to those copies, and it is a public record that the plaintiff can point to for as long as the copies last.
Choose the defendant with care. The author is the defendant in the ordinary case, because the author controls the accounts and the judgment against the author is what the platforms act on. A claim against the platform itself is possible, and Thorpe shows that a host on notice may be answerable for what it leaves up (¶ 146), but it must be properly served, in the United States if that is where it is, and the orders against it must be narrow and supported by evidence (Braet at ¶¶ 75 and 80; Obsidian at ¶ 7). That is a different and longer proceeding, and it is not the way the content usually comes down.
How we can help
Grigoras Law acts for individuals and businesses in Toronto and across Ontario who have been defamed online, from the demand letter through the judgment and its delivery to the platforms, and for defendants who have been served with a claim or an injunction motion. We draft the orders in the form the platforms act on, and we deal with the republications and the anonymous accounts that follow. Our defamation practice and our reputation management practice are described on the pages linked here, the hub page on defamation and reputation management covers the group as a whole, and the page on social media and internet law covers the platforms themselves. For an interlocutory injunction before judgment, see our page on urgent commercial remedies.
This article is general information about the law of Ontario as of October 2026 and is not legal advice. Every case turns on the words complained of, the platform, the evidence and the defendant, and the form of the order decides whether the content comes down, so the steps described here deserve advice before the first demand is sent.
This article is part of the firm’s work in defamation.