An Analysis of Latham v Greenwich [2026] FCAFC 82
1. Introduction
On 10 June 2026 the Full Court of the Federal Court of Australia (Colvin, Wheelahan and Abraham JJ) delivered judgment in Latham v Greenwich [2026] FCAFC 82, dismissing both an appeal and a cross-appeal from the decision of O’Callaghan J in Greenwich v Latham [2024] FCA 1050. The primary judge had held Mr Mark Latham, a member of the New South Wales Legislative Council, liable in defamation to Mr Alexander Greenwich, the member of the Legislative Assembly for Sydney, in respect of a single post on Twitter (now X), and had assessed damages, including aggravated damages, in the sum of $140,000. The Full Court found no error at any point in the primary judge’s reasoning.
The decision warrants attention well beyond the immediate parties because it is the most comprehensive appellate treatment to date of the serious harm element introduced into the uniform defamation legislation by the 2020 reform package – in this case, s 10A of the Defamation Act 2005 (NSW). The joint reasons of Wheelahan and Abraham JJ, with whom Colvin J relevantly agreed, resolve three questions of immediate practical significance: how the expression “serious harm” is to be construed; whether the element attaches to the publication of the defamatory matter or instead to each defamatory imputation; and how serious harm may be proved, including whether a claimant must adduce direct evidence comparing reputation before and after publication.
In answering the first of those questions the Full Court declined to follow the approach of the majority of the Court of Appeal of the Australian Capital Territory in Supaphien v Chaiyabarn [2026] ACTCA 5 and endorsed the line of authority culminating in MG v PJ [2025] QCA 99: at [59]–[63]. There is now an open divergence between intermediate appellate courts on the construction of the Australian serious harm provisions which only the High Court can finally resolve. The judgment also contains significant guidance on the statutory defence of honest opinion, on the common law qualified privilege for a reply to an attack (including whether “proportionality” is an element of that defence), on the post-2020 architecture for awards of aggravated damages, and on the discipline expected of appellants in framing and maintaining grounds of appeal.
The decision has a particular resonance for practitioners in Western Australia, although for a distinctive reason: Western Australia has not adopted the serious harm element, and the Defamation Act 2005 (WA) continues to operate in its original form, retaining the statutory triviality defence. The element nonetheless arises regularly in Western Australian practice. Online publications are commonly read in more than one Australian jurisdictional area, engaging the choice of law rule in s 11 of the uniform Acts, under which the substantive law of the jurisdictional area with the closest connection to the harm governs; and defamation claims concerning publications with a national reach are frequently brought in the Federal Court. The Full Court’s framework will also inform any future adoption of the element in this State.
2. Relevant Legal Framework
2.1 The serious harm element
At common law, upon proof of the publication of defamatory matter, damage to reputation was presumed. The Model Defamation Amendment Provisions 2020, given effect in New South Wales by the Defamation Amendment Act 2020 (NSW) from 1 July 2021, altered that position in the adopting jurisdictions. Section 10A(1) of the Defamation Act 2005 (NSW), set out by the Full Court at [52], provides:
It is an element (the serious harm element) of a cause of action for defamation that the publication of defamatory matter about a person has caused, or is likely to cause, serious harm to the reputation of the person.
The same reform repealed the statutory defence of triviality, formerly contained in s 33, which had been engaged if the defendant proved that the circumstances of publication were such that the plaintiff was unlikely to sustain any harm: at [55] and [58]. Under s 10A(5), if a party applies, the serious harm element is to be determined as soon as practicable before trial unless special circumstances justify postponement: at [58]. The extrinsic materials show that the element was introduced to weed out trivial, spurious, frivolous and vexatious claims: at [58], adopting the account of the legislative history in Mond v The Age Company Pty Ltd [2025] FCA 442 at [373]–[377].
2.2 The United Kingdom lineage
Section 10A bears some resemblance to s 1(1) of the Defamation Act 2013 (UK), but the Full Court emphasised that the two provisions are in different terms and have different legislative and common law contexts: at [53]. The United Kingdom provision was enacted against the background of the abuse of process jurisdiction recognised in Jameel (Yousef) v Dow Jones & Co Inc [2005] EWCA Civ 75; QB 946 and the common law threshold of substantiality identified in Thornton v Telegraph Media Group Ltd [2011] 1 WLR 1985 at [96], and it was held in Lachaux v Independent Print Ltd [2019] UKSC 27; [2020] AC 612 at [12] to raise the threshold of seriousness above that envisaged in those cases: at [54]. The anterior Australian setting was different: it was the availability, under the uniform legislation, of the statutory triviality defence: at [55]. The Full Court accepted that the New South Wales provision “might have been inspired by” the United Kingdom provision, but rejected the contention that it was modelled on it: at [62].
2.3 Australian authority before the appeal
The serious harm element had been considered at first instance in the Federal Court in Russell v Australian Broadcasting Corporation (No 3) [2023] FCA 1223; 303 FCR 372 at [252]–[254], Deeming v Pesutto (No 3) [2024] FCA 1430, Mond and Griniezakis v Morelas (Trial Judgment) [2026] FCA 156, and by the Full Court in Selkirk v Wyatt [2024] FCAFC 48; 302 FCR 541: at [56].
The fault line in the authorities derived from Rader v Haines [2022] NSWCA 198, a case governed by the United Kingdom Act, in which Brereton JA stated at [27] that “‘serious’ harm sits on the spectrum above ‘substantial’ but below ‘grave’”. Basten AJA, writing separately, expressed a reservation at [91] about the risk in seeking to place the statutory term on a scale between other terms of equal imprecision. The reservation was subsequently preferred in Selkirk v Hocking (No 2) [2023] FCA 1085 at [30], Peros v Nationwide News Pty Ltd (No 3) [2024] QSC 192 at [95], Mond at [374], MG v PJ [2025] QCA 99 at [83] and Moore v Martin [2026] NSWSC 493 at [28]: collected at [61]. By contrast, in Supaphien v Chaiyabarn [2026] ACTCA 5 a majority of the ACT Court of Appeal (McCallum CJ and Loukas-Karlsson J) applied the reasoning of Brereton JA in Rader v Haines to the corresponding provision in s 122A(1) of the Civil Law (Wrongs) Act 2002 (ACT), concluding at [73] that the legislation intended to introduce a higher threshold than non-trivial harm: discussed at [62]. That was the state of the authorities when the appeal was heard.
3. The Facts of the Case
3.1 The parties and the publications
Mr Latham and Mr Greenwich are both members of the New South Wales Parliament and may be described as political opponents. Mr Greenwich is openly homosexual and is an advocate for members of the LGBTQIA+ community; both men sit as independents, Mr Latham having formerly been a member of the One Nation party: at [18]. Both campaigned in, and were re-elected at, the State election held on 25 March 2023: at [19].
Two earlier publications place the matter complained of in context. On 22 March 2023 the Sydney Morning Herald website republished a statement that Mr Greenwich had given to a journalist: at [19] –
Mark Latham is a disgusting human being and people who are considering voting for One Nation need to realise they are voting for an extremely hateful and dangerous individual who risks causing a great deal of damage to our state.
On the same day Ms Susan Metcalfe published a tweet quoting that statement, adding Mr Greenwich’s name and a video captioned “Video shows LGBTQ protesters pleading for help outside Mark Latham event”: at [20]. About a week later, on 30 March 2023, Mr Latham posted the publication for which he was held liable (the primary tweet) in response to the Metcalfe tweet. The words complained of were: at [21] –
Disgusting?
How does that compare with sticking your dick up a bloke’s arse and covering it with shit?
Mr Greenwich relied on two imputations: first, arising on the natural and ordinary meaning of the primary tweet, that he “engages in disgusting sexual activities”; and secondly, by way of true innuendo, that he “is not a fit and proper person to be a member of the NSW Parliament because he engages in disgusting sexual activities”, the extrinsic facts being knowledge of the Metcalfe tweet and of his membership of Parliament: at [22]–[24].
3.2 The extent of publication and the response
There were 6,171 direct views of the primary tweet before Mr Latham deleted it, two hours and twenty minutes after posting: at [68]. A television journalist, Mr Hugh Riminton, republished it by reposting it on Twitter, where it was viewed 654,700 times: at [69]. Further responses followed mainstream media reporting of the controversy, together with feedback and emails to Mr Greenwich’s electorate office: at [69]. The primary judge characterised the online responses as unleashing “hate-filled venom” and found that the communications Mr Greenwich received as a result of the primary tweet clearly indicated harm to his reputation: J [188] and [196], described at [70]. The publication provoked what the Full Court described as a torrent of vile, threatening, abusive and deranged communications, including voicemail messages that were “particularly menacing, and very disturbing”: at [138]–[139].
3.3 The trial and the appeal
The primary judge held that the primary tweet conveyed the first imputation and was defamatory; that the second imputation was not conveyed; that the serious harm element in s 10A was satisfied; and that defences of statutory honest opinion under s 31 and common law qualified privilege (reply to attack) failed. His Honour assessed general damages at $100,000 and made a separate award of aggravated damages of $40,000: at [25] and [140]. A claim for a permanent injunction was later dismissed: Greenwich v Latham (No 2) [2025] FCA 131, noted at [17]. Mr Greenwich’s solicitors had sent a concerns notice dated 19 April 2023 calling for a reasonable apology, which was rejected; no apology was ever made: J [274], summarised at [151].
Mr Latham’s appeal challenged the findings on meaning, defamatory quality, serious harm, both defences, and both damages awards; Mr Greenwich cross-appealed the rejection of the true innuendo: at [26]–[27]. The appeal and the cross-appeal were each dismissed: at [165].
4. Analysis of the Court’s Reasoning
4.1 Meaning and defamatory quality: impression, not analysis
The question whether the primary tweet conveyed the first imputation attracted the correctness standard of appellate review, the primary judge having enjoyed no relevant advantage: at [32]. The joint reasons emphasised the distinction between the capacity of a matter to convey a defamatory meaning and the factual question whether it did so, observing that authorities such as Favell v Queensland Newspapers Pty Ltd [2005] HCA 52; 221 ALR 186 and Trkulja v Google LLC [2018] HCA 25; 263 CLR 149 were concerned with capacity, and that Sedley LJ’s aphorism in Berezovsky v Forbes Inc [2001] EWCA Civ 1251; EMLR 45 at [16] that the exercise is one “in generosity and not parsimony” is directed to capacity rather than to the meaning in fact conveyed: at [34]. Colvin J’s separate reasons develop the same distinction: at [3]–[5].
As to the meaning in fact conveyed, the ordinary reasonable reader of a social media post has the typical attributes of a social media user: Stocker v Stocker [2019] UKSC 17; [2020] AC 593 at [41]–[43]; Bazzi v Dutton [2022] FCAFC 84; 289 FCR 1 at [29], cited at [36]. Twitter is a casual, conversational medium; it is the impression conveyed by a post, not an interpretation that is the product of deep analysis, that determines the objective single meaning: at [36]. Mr Latham’s submissions – that the tweet referred only to one specific form of intercourse and conveyed no more than that Mr Greenwich engaged in homosexual male intercourse – “rested upon analysis rather than impression”: at [37]. The ordinary reasonable reader would quickly have been left with the impression that Mr Greenwich engages in disgusting sexual acts, a meaning squarely within the first imputation: at [37].
Colvin J characterised the tweet as “an aggressive invocation of the kind of trope that has been used to foment hatred and violence towards homosexual men”, but was careful to identify why it was defamatory rather than merely insulting: the tweet connected the notion of being a disgusting human being with a profane description of sexual activity, conveying that aspects of Mr Greenwich’s sexual behaviour were properly characterised as nauseating; applying the standards of ordinary reasonable people, statements of that kind make such people think less of the person: at [11]–[13]. His Honour rejected the contention that the tweet was a classic tu quoque response, describing it instead as “a blatant attempt to denigrate by making an unrelated statement about Mr Greenwich’s sexual activities”: at [12].
The test for defamatory quality remained that of general community standards: whether the matter would be likely to cause ordinary reasonable people to think less of the claimant: Reader’s Digest Services Pty Ltd v Lamb [1982] HCA 4; 150 CLR 500 at 507; Radio 2UE Sydney Pty Ltd v Chesterton [2009] HCA 16; 238 CLR 460 at [36]; Harbour Radio Pty Ltd v Trad [2012] HCA 44; 247 CLR 31 at [54], cited at [44]. Mr Latham’s submissions failed because they addressed a different, blander meaning – that Mr Greenwich was a homosexual man who engaged in sexual activity with another male – and did not meet the sting of the meaning actually conveyed, with its graphic reference to activities involving faeces: at [51]. The cross-appeal failed for a complementary reason: the ordinary reasonable reader, reading impressionistically, would not pause, consider the extrinsic facts and “reason tangentially” towards the second imputation concerning fitness for office: at [43].
4.2 The construction of “serious harm”
The term “serious harm” is not defined. It is, the Full Court held, an open textured phrase to be construed by reference to text, context and purpose in accordance with ordinary principles of statutory construction: at [57]. The purpose disclosed by the extrinsic materials – to weed out trivial, spurious, frivolous and vexatious claims – and the design of s 10A(5), which assumes that the element may be capable of ready determination before trial, together indicate that “serious harm to reputation” is harm that is at least not trivial or minor, rather than “an especially high threshold requiring extensive evidence”: at [58]. At the same time, serious harm is not simply the converse of the repealed triviality defence: proof of some harm to reputation is not to be equated with harm “of a higher order warranting the characterisation of ‘serious’”: at [58], citing Peros at [89].
The Full Court squarely rejected the calibrated “spectrum” approach derived from Brereton JA’s judgment in Rader v Haines. Agreeing with Basten AJA’s reservation, their Honours invoked the High Court’s warnings against straying from legislative text and deploying judicial formulations as a substitute for it: Baini v The Queen [2012] HCA 59; 246 CLR 469 at [14]; Shi v Migration Agents Registration Authority [2008] HCA 31; 235 CLR 286 at [92]: at [60]. Because Rader v Haines construed the United Kingdom Act, it carried no special precedential status of the kind recognised in Australian Securities Commission v Marlborough Gold Mines Ltd [1993] HCA 15; 177 CLR 485 and Hill v Zuda Pty Ltd [2022] HCA 21; 275 CLR 24 at [26]: at [61]. And while the majority in Supaphien had applied the Rader v Haines formulation on the footing that the Australian provisions were modelled on the United Kingdom provision, the Full Court did not agree that they were; the decisions preferring Basten AJA’s approach, including the Queensland Court of Appeal’s decision in MG v PJ, “should be followed”: at [62].
The correct approach is that identified in MG v PJ at [83]–[84]: the plain and ordinary meaning of the word “serious” is to be adopted, having regard to the purpose of the provision. The result is that the term calls for the court “to reach an evaluative conclusion on the evidence in the exercise of its own judgment”, accepting that precision will rarely be possible: at [63], citing Mond at [407]–[408] and Lachaux at [21]. Two further observations frame the evaluation. Damage to reputation is not a commodity, and every defamation is unique: Rogers v Nationwide News Pty Ltd [2003] HCA 52; 216 CLR 327 at [66], cited at [63]. And harm to reputation is not like a bodily injury: reputation exists in the minds of others and may not be homogeneous, so the impact of the matter on those to whom it was published may result in serious harm without the impact being uniform: at [63].
4.3 The element attaches to the matter, not to each imputation
Mr Latham submitted that serious harm must be proved to stem from publication of the imputation, and that few of the identified communications related to the imputation as opposed to the primary tweet. The Full Court rejected the premise, observing that there is on occasion “a tendency for some parties to defamation proceedings to make submissions as if the Defamation Act 1974 (NSW), under which each imputation was a separate cause of action, had not been repealed”: at [80]. Under s 8 of the current Act a person has a single cause of action in relation to the publication of defamatory matter even if more than one defamatory imputation is conveyed, and s 6(3) directs that the general law applies as if the 1974 and 1958 Acts had never been enacted: at [80].
Their Honours drew support from Warby J’s holding in Sube v News Group Newspapers Ltd [2018] 1 WLR 5767 at [27] that the statutory test is not phrased in terms of the reputational harm caused by a meaning or an imputation. Addressing the apparently contrary statement of Warby LJ in Blake v Fox [2025] EWCA Civ 1321 at [113] – that the claimant must show serious harm “in relation to each imputation” – the Full Court observed, having traced the authorities his Lordship cited: “We cannot help but think that there must have been a slip in Blake v Fox at [113], and that the reference to ‘each imputation’ should be understood as a reference to ‘each statement’”: at [81].
The holding is that s 10A(1) is directed to the publication of the defamatory matter, not to a defamatory meaning or imputation; a field of inquiry connecting individual imputations to serious harm “is not contemplated by the terms of s 10A and would be difficult where two or more imputations are found to be conveyed by the one matter”: at [82]. On the facts the point was in any event immaterial, because there was a strong correlation between the publication in its defamatory sense and the reactions the subject of the primary judge’s findings: at [82].
4.4 Proof by inference: the eight considerations and the evidentiary onus
Ordinary principles apply to the sufficiency of proof of actual harm to reputation, with allowance for the inherent difficulty of marshalling direct evidence: harm to reputation may lie lurking in unknown places: at [64], reflecting Lord Hailsham’s observation in Cassell & Co Ltd v Broome [1972] AC 1027 at 1071. Proof of serious harm “can and often will depend upon drawing inferences from all the circumstances”: at [64], citing Lachaux at [21]. The authorities show that the circumstances from which the inference may be drawn include a combination of some of the following considerations, set out at [64]:
(a) the inherent tendency of the defamatory matter and its actual or probable impact on those to whom it was published – the gravity of the imputations;
(b) the extent of publication, any likely republication, and any likely “grapevine effect” or “percolation”;
(c) whether there was a prompt apology, and the breadth of the audience to which any apology was published;
(d) the reputation of the publisher – a publication by a generally reputable publisher represented as reliable and the product of research might be more capable of causing serious harm than one by a publisher that is not so reputable;
(e) the medium of the publication – a transient publication might cause less harm than one in print or available online for an extended period or indefinitely;
(f) the situation of the applicant, including the applicant’s existing reputation in the relevant sector;
(g) the identity of the persons to whom the defamatory matter was published; and
(h) the inherent probabilities generally.
This is “not a check list, and it is not exhaustive”; in some cases a consideration of only some circumstances may readily lead one way or the other: at [65], citing Mond at [371]. Although the element is concerned with actual rather than presumed harm, in a case where causation of serious harm from widespread publication is a matter of inference, the single meaning the publication is found to bear “is a sound starting point” in determining the probable impact on those to whom it was published: at [66].
Applying that framework, the Full Court rejected the submission that proof of the element will almost always require “before and after” reputation evidence, insofar as that submission asserted a requirement of direct comparison evidence: at [78]. Mr Greenwich, who enjoyed an unchallenged prior good reputation, established serious harm “as an inference arising from all the circumstances, central elements of which were the vile social media and other responses that the primary tweet and its republication provoked”: at [78]. Significantly for defendants, their Honours held that where such an inference otherwise arises, a respondent who wishes to contest it – for example on the ground of a pre-existing relevant bad reputation or some unusual feature rebutting the inference – bears an evidentiary onus to put the question in issue, although the legal onus remains throughout on the claimant: at [78], citing Purkess v Crittenden (1965) 114 CLR 164 at 167–168.
The Court accepted that the element requires evidence that people thought less of the applicant because of the publication, but reiterated that the conclusion “can and often will be an inference arising from all the circumstances”: at [79]. Mr Latham’s speculation that those who responded harboured “well-entrenched perceptions” of Mr Greenwich – drawing on an observation in Palmer v McGowan (No 5) [2022] FCA 893; 404 ALR 621 – was no answer, because no body of evidence materially undermining the inference was adduced: at [79]. Nor did the findings of continuing good reputation “across the chamber” assist: reputation exists in the minds of others and harm need not be uniform; “an inference of reputational harm in the minds of some persons to whom the matter is published may be sufficient to support a finding of serious harm”: at [83]. Colvin J added that the evidence that the tweet fomented hatred and vitriol did not establish that it operated only upon those who already harboured prejudice; it was open to infer that ordinary reasonable people formed adverse views of Mr Greenwich’s reputation: at [16].
4.5 Honest opinion: the defence fails for want of proper material
Section 31(1) of the Defamation Act 2005 (NSW) provides a defence where the matter was an expression of opinion rather than a statement of fact, the opinion related to a matter of public interest, and the opinion was based on proper material as defined in s 31(5); the defence is defeated only if the claimant proves the opinion was not honestly held: s 31(4)(a), set out at [84]–[86]. Mr Latham’s pleading did not identify in terms what the expression of opinion was – an omission the Full Court described as important, because the section fastens upon the distinction between an expression of opinion and a statement of fact: at [87]–[88], citing Massoud v Nationwide News Pty Ltd [2022] NSWCA 150; 109 NSWLR 468 at [195]–[197]. The primary judge assumed in Mr Latham’s favour that the tweet was an expression of opinion, identified (adopting the claimant’s formulation) as an opinion that Mr Greenwich was disgusting because of the act the tweet described: at [88].
The difficulty was proper material. Mr Latham had pleaded, but abandoned at trial, the particular that Mr Greenwich “is an openly gay man who has participated in homosexual sexual activities”; the consequence was that there was no pleading and no evidence that Mr Greenwich engaged in the act described in the primary tweet: at [91] and [95]. The Full Court held that the only candidate for proper material to support the opinion captured by the word “disgusting” was the description of the sex act in the tweet itself; the ordinary reasonable reader would consider the opinion to be based on those facts; and no attempt had been made to establish their substantial truth: at [108]. The defence therefore failed without any need to address the public interest element or the other submissions: at [108]. The agreed background facts – the attack, the church forum incident, the parties’ public positions – could not assist, because the opinion the reader would identify was not based on them.
4.6 Reply to attack: the response must be commensurate with the occasion
The defence of common law qualified privilege requires the occasion of the privilege to be identified, because the nature of the occasion informs whether the publication was relevant to it and whether it was made for a foreign purpose: at [110]. Mr Latham relied on the recognised reply-to-attack privilege, which accommodates robust answers, defences and counterattacks relevant to the occasion: at [111]. The authorities tolerate vigour – “[i]f you are attacked by a prize fighter you are not bound to adhere to the Queensberry rules in your defence”: Turner v Metro-Goldwyn-Mayer Pictures Ltd [1950] 1 All ER 449 at 471 (Lord Oaksey), cited at [111] – but the use of language exceeding the occasion is significant in two distinct ways: at [112]–[113]. First, the words may fall outside the privileged occasion altogether because they are not relevant to it: Cush v Dillon [2011] HCA 30; 243 CLR 298 at [19] and [25]; Adam v Ward [1917] AC 309 at 320–321. A response must be relevant to the attack and “be commensurate with” the occasion: Penton v Calwell (1945) 70 CLR 219 at 234 (Dixon J); Harbour Radio at [27], [33] and [35]. Secondly, where the words fall within the occasion, exaggerated language may be evidence of malice: Cush v Dillon at [24]–[25].
The defence failed at the first step. The primary tweet was “only superficially germane to the attack in the sense that both publications used the word ‘disgusting’”; as a matter of substance it “was not a reply at all”, because the imputation it conveyed was not germane to an attack concerning Mr Latham’s fitness for office; it was “retaliation on a different front”: at [131]. Disparaging Mr Greenwich on the basis of a disgusting sexual act was not a relevant response and was therefore not commensurate with the occasion: at [131].
Of particular interest is the Full Court’s treatment of “proportionality”. Their Honours found force in the submission that no High Court authority recognises a separate requirement that a response be proportionate to the attack before the occasion of privilege arises: at [132]. The proportionality line in the Australian cases runs through Gould v Jordan (No 2) [2021] FCA 1289 at [68] and [71], which drew on Western Australian authority: Accommodation West Pty Ltd v Aikman [2017] WASC 157 at [191] and Harding v Essey [2005] WASCA 30; 30 WAR 1 at [10]; see also Heytesbury Holdings Pty Ltd v City of Subiaco (1998) 19 WAR 440 at 459. The Full Court noted that proportionality in this context appears in English law where rights under the Human Rights Act 1998 (UK) are engaged, legislation which does not affect the law of New South Wales or the common law of Australia: at [132]. Having upheld the rejection of the privilege on the commensurateness ground, the Court found it unnecessary to decide whether any independent proportionality requirement exists, while observing that a disproportionate response may be evidence of malice: at [133]. The question of malice itself was “swamped” by the conclusion that the occasion was not established – to entertain it would be “to run up a dry gully”: at [134]. The submissions invoking the implied freedom of political communication recognised in Lange v Australian Broadcasting Corporation [1997] HCA 25; 189 CLR 520 were not addressed: the electoral privilege considered in Roberts v Bass [2002] HCA 57; 212 CLR 1 had not been run at trial: at [114] and [135].
4.7 Damages and the conduct of the appeal
The award of $100,000 in general damages was upheld. The assessment had to serve the three purposes identified in Carson v John Fairfax & Sons Ltd [1993] HCA 31; 178 CLR 44 at 59–60 – consolation for personal distress and hurt, reparation for harm to reputation, and vindication: at [147]. The assessment of damages for non-economic loss is an intuitive process; absent specific error, a complaint of manifest excess invokes the final basis for appellate intervention in House v The King [1936] HCA 40; 55 CLR 499 at 504–505: at [146], applying Nationwide News Pty Ltd v Rush [2020] FCAFC 115; 380 ALR 432 at [472]–[479]. A submission that hurt to feelings had to be traced to the imputation rather than the publication suffered from the same flawed premise addressed at [80] and, in any event, cut across the concession at trial that Mr Greenwich was significantly hurt by the primary tweet: at [145].
The award of aggravated damages engaged the post-2020 statutory architecture. At common law aggravated damages were not a separate head of damage; aggravating conduct and the increased harm it caused were taken into account in the assessment of general compensatory damages: at [148], citing Mond at [501]. Section 35(1) now imposes a cap on damages for non-economic loss – $443,000 at the time of trial – to be awarded only in a most serious case: at [149]. By ss 35(2A) and (2B), the cap does not limit the power to award aggravated damages where warranted, and such an award “is to be made separately” from the award for non-economic loss: at [149]–[150]. This may be contrasted with the position under s 35 as originally enacted, where a finding warranting aggravated damages simply uncapped the global award: Bauer Media Pty Ltd v Wilson (No 2) [2018] VSCA 154; 56 VR 674; Rush, discussed at [150]. The conduct relied on must be improper, unjustifiable or lacking in bona fides: Triggell v Pheeney (1951) 82 CLR 497 at 514, cited at [151]. Here the aggravating matters included the absence of any proper basis for the “slur”, the rejection of the concerns notice, the absence of any apology in circumstances where one was self-evidently called for, a false public assertion that an apology had been made, and a course of “doubling down” tweets and a radio interview – conduct which the primary judge held “rubbed salt in the wound” caused by the primary tweet, reflecting the language of Lord Donaldson MR in Sutcliffe v Pressdram Ltd [1991] 1 QB 153 at 170: J [274] and [282], summarised at [151]–[155].
Finally, the Court refused leave to amend the notice of appeal to accommodate a substantially recast challenge to the aggravated damages award, advanced in a 13-page submission prepared overnight during the hearing. The ground as filed – that the primary judge “erred in awarding the Respondent aggravated damages” – was “completely unfocused”; the new points, including pleading objections, had not been taken at trial; and expansion of the appeal would have contravened the overarching purpose in s 37M of the Federal Court of Australia Act 1976 (Cth): at [156]–[162], citing SunshineLoans Pty Ltd v Australian Securities and Investments Commission [2025] FCAFC 34; 308 FCR 474 at [131]–[134]. Their Honours observed that trial counsel had plainly appreciated that “the best advocacy is selective and economical”: at [163], citing Whisprun Pty Ltd v Dixon [2003] HCA 48; 200 ALR 447 at [18].
5. Assessing the Consequences and Quantification
The monetary outcome bears analysis. The general damages award of $100,000 represented somewhat under a quarter of the then applicable cap of $443,000: at [142] and [149]. For a single tweet, deleted within two hours and twenty minutes, that is a substantial sum. The quantum was driven by the extent of republication (654,700 views of the Riminton repost alone), the gravity and graphic character of the imputation, the largely unchallenged evidence of significant hurt to feelings, the menacing responses the publication provoked, and the absence of any genuine apology: at [136]–[139] and [147].
The separate award of $40,000 in aggravated damages, described by the primary judge as modest, is significant chiefly for its mechanics. The Full Court’s analysis confirms that in the amended jurisdictions aggravated damages now sit wholly outside the s 35(1) cap and must be quantified separately: at [149]–[150]. Practitioners should plead and particularise aggravation distinctly, and should recognise that post-publication conduct – the response to a concerns notice, the presence or absence of an apology, continued public commentary – may convert a contained damages exposure into a larger one. Mr Latham’s post-publication tweets and interview added $40,000 to the judgment against him.
The decision also affects how serious harm disputes will be litigated. Because the element may be established inferentially from the matrix of circumstances identified at [64], the evidentiary contest will frequently centre on publication analytics, republication chains and the content of audience responses. Claimants will need to capture that material early, before posts and metrics disappear; publishers should appreciate that prompt deletion, while relevant to the extent of publication, did not avoid liability here, and that an early, adequately published apology engages consideration (c) directly. The early determination mechanism in s 10A(5) remains available, but a respondent who seeks early determination on facts involving widespread hostile reaction invites an adverse finding at an interlocutory stage, with costs.
As to the costs of the appeal itself, the Court ordered the parties to confer and provide an agreed minute, failing which costs will be determined on the papers: order 3 and at [166].
6. Worked Example
Consider a hypothetical. A strata chairperson in New South Wales posts in a public community Facebook group of 12,000 members about a local building contractor: “Warning to everyone – these people are crooks. They take deposits from pensioners and never come back. Disgraceful.” Group analytics record 4,200 views. Sixty-three comments follow, many abusive of the contractor (“scum”, “should be in jail”), and screenshots are reposted to a trade forum. The contractor, who had no prior adverse publicity, loses three booked jobs in the following fortnight. The post is deleted after two days; no apology is made.
For the contractor (the claimant), the analysis under Latham v Greenwich proceeds inferentially through the considerations at [64]. The imputation – dishonest exploitation of vulnerable customers – is grave (consideration (a)). Publication was substantial within the relevant local market, with proved republication and an obvious grapevine effect among prospective customers (considerations (b) and (g)). There was no apology (consideration (c)). The post, while ultimately deleted, remained available for two days and survives in screenshots (consideration (e)). The contractor’s livelihood depends on reputation in the precise sector in which the matter percolated (consideration (f)). The hostile comments are themselves evidence that recipients thought less of the contractor: at [79] and [83]. The lost bookings provide direct confirmation, but on the authority of Latham v Greenwich the claim would not fail without them: serious harm may rest on inference alone: at [78].
For the chairperson (the respondent), the lessons are equally clear. Speculation that group members already held entrenched views about the contractor will achieve nothing; the evidentiary onus identified at [78] requires actual evidence of a pre-existing relevant bad reputation, such as prior complaints or adverse coverage, to displace the inference. Supportive comments defending the contractor may be marshalled to qualify the harm. Deletion after two days mitigates but does not answer the claim. An early s 10A(5) application to determine serious harm would be unwise on these facts. The respondent’s better strategy lies elsewhere: a prompt, prominent apology published to the same group (consideration (c)), an offer to make amends, and attention to substantive defences – honest opinion will require identification of the opinion and proof of the substantial truth of its factual foundation, which Latham v Greenwich at [108] shows will fail if the foundational facts cannot be proved.
Transposed to Western Australia, the analysis changes at the threshold. If the publication were read substantially only in this State, the substantive law of Western Australia would apply under s 11 of the uniform Acts, and there would be no serious harm element to prove. The contractor’s cause of action would be complete upon publication of the defamatory matter, with damage presumed; the chairperson would bear the onus of any triviality defence under s 33 of the Defamation Act 2005 (WA). Where the readership straddles jurisdictions – the ordinary case for social media – the closest connection test determines which regime governs, and the difference between the regimes may be outcome-determinative for marginal claims.
7. Practitioner Guidance: A Step-by-Step Framework
Step 1 – Identify the applicable substantive law before pleading
Determine where the matter was published in the relevant sense and apply s 11 of the uniform Acts. The serious harm element applies in the jurisdictions that adopted the 2020 amendments; it does not apply in Western Australia. For multi-jurisdictional online publications, identify the Australian jurisdictional area with the closest connection to the harm. The forum (including the Federal Court) does not alter the applicable substantive law.
Step 2 – Plead serious harm by reference to the matter, not each imputation
Frame the serious harm case on the footing that s 10A(1) attaches to the publication of the defamatory matter as a whole: at [80]–[82]. Resist, and meet, submissions that parse serious harm imputation-by-imputation; the Full Court has identified such submissions as a relic of the repealed 1974 Act regime: at [80].
Step 3 – Build the inferential matrix early
Collect and preserve, at the outset of the retainer, the material from which the [64] considerations can be proved: platform analytics and view counts, the republication chain, screenshots with metadata, the content of comments and messages, and evidence of the claimant’s standing in the relevant sector. This material disappears quickly; a preservation letter should issue immediately.
Step 4 – Treat hostile audience reaction as evidence, not noise
Abusive and derisive responses provoked by the publication are themselves capable of supporting the inference that recipients thought less of the claimant: at [79] and [83]. Capture them. Evidence of hurt to feelings serves the damages case; evidence of audience reaction serves the serious harm case; keep the two streams distinct in proofing and in submissions.
Step 5 – For respondents, answer the inference with evidence
Once the circumstances support an inference of serious harm, a respondent bears an evidentiary onus to put rebutting matters in issue – a pre-existing relevant bad reputation, a break in causation, or some unusual feature: at [78], applying Purkess v Crittenden. Speculative assertions about “well-entrenched perceptions” will not do: at [79]. If rebutting evidence does not exist, advise the client accordingly and direct attention to defences and to mitigation, including an early apology.
Step 6 – Consider the timing of a serious harm determination
Section 10A(5) requires determination of the element before trial on application, absent special circumstances: at [58]. An early application suits a respondent facing a genuinely trivial claim; it is dangerous where the audience reaction was substantial. Claimants resisting early determination should be ready to point to the inferential matrix rather than promising direct evidence at trial.
Step 7 – Plead defences with precision
For honest opinion, identify in the pleading the opinion said to have been expressed, the matter of public interest to which it relates, and the proper material on which it was based, and be in a position to prove that material substantially true so far as it is factual: at [88] and [108]. For reply-to-attack privilege, identify the occasion precisely and test the proposed response for commensurateness: a counterattack that does not engage with the substance of the attack is “retaliation on a different front” and will fall outside the occasion: at [131]. Treat proportionality as unresolved but assume a disproportionate response will be deployed as evidence of malice: at [132]–[133].
Step 8 – Preserve points at trial and frame focused grounds of appeal
Points not taken at trial – including pleading objections – will rarely be permitted to ground an appeal, and unfocused grounds will confine the appellant to the written submissions filed: at [157]–[163]. Draft grounds of appeal that identify the asserted error with specificity, and ensure that trial counsel put every responsive submission that the client may later wish to maintain.
8. Evidence and Arguments Available to Each Side
8.1 For the claimant
The claimant’s armoury after Latham v Greenwich centres on circumstantial proof: platform analytics establishing views and reach (6,171 direct views and 654,700 republication views were decisive here: at [68]–[69]); the republication chain and grapevine percolation; the content and tone of audience responses, deployed as evidence that recipients thought less of the claimant: at [79] and [83]; the gravity of the single meaning as the starting point for probable impact: at [66]; evidence of prior good reputation in the relevant sector; the absence or refusal of an apology, which serves double duty under consideration (c) and as aggravation where a concerns notice was rebuffed: J [274], at [151]; and unchallenged evidence of hurt to feelings in support of general damages. Where the respondent has continued to publish, each item of “doubling down” conduct should be proved and particularised as aggravation: at [151]–[155].
8.2 For the respondent
The respondent’s realistic avenues are these: evidence (not assertion) of a pre-existing relevant bad reputation or of audience predisposition sufficient to rebut the inference of serious harm: at [78]–[79]; evidence of supportive or sympathetic responses qualifying the asserted harm: at [71]; a causation analysis isolating harm attributable to other publications not sued upon, noting that a concession of causation at trial will be decisive: at [73] and [83]; reliance on the transience of the medium, prompt deletion and limited audience under considerations (b) and (e); a prompt and adequately published apology; a properly pleaded honest opinion defence with provable proper material; and a reply-to-attack privilege confined to a response that genuinely engages the substance of the attack. At trial, any challenge to the cause of the claimant’s hurt feelings must actually be put: an appellate court will hold the respondent to the way the trial was conducted: at [145]. Finally, the respondent should weigh the forensic cost of pressing meaning submissions that recast the publication into something blander than its sting; both Colvin J at [14] and the joint reasons at [51] treated that strategy as self-defeating.
9. Key Takeaways for Legal Practice
1. “Serious harm” bears its ordinary meaning, construed in light of the statutory purpose. There is no judicially calibrated spectrum on which “serious” sits above “substantial” and below “grave”. Submissions built on the Rader v Haines formulation should not be advanced in the Federal Court: at [59]–[63].
2. Intermediate appellate authority is divided, and the Full Court has chosen its side. Supaphien v Chaiyabarn applies the spectrum approach in the Australian Capital Territory; MG v PJ, Moore v Martin and now Latham v Greenwich reject it. Until the High Court intervenes, the governing approach will depend on the court in which the proceeding is brought, a matter to weigh in forum selection: at [61]–[62].
3. The serious harm element attaches to the publication of the matter, not to each imputation. Pleadings and submissions should connect the harm to the publication as a whole. The contrary approach is a relic of the Defamation Act 1974 (NSW): at [80]–[82].
4. Serious harm may be proved wholly by inference. There is no requirement of direct “before and after” comparison evidence. The eight considerations at [64] supply the working framework; the single meaning is the starting point for probable impact: at [64]–[66] and [78].
5. Once the inference arises, the respondent bears an evidentiary onus. A respondent who wishes to contest serious harm must adduce evidence putting the rebutting matter in issue; speculation about audience predispositions is insufficient: at [78]–[79], applying Purkess v Crittenden.
6. Hostile abuse provoked by a publication is evidence of reputational harm. Harm to reputation need not be uniform across the audience; an inference of harm in the minds of some recipients may suffice, even where the claimant retains broad support: at [83]; Colvin J at [16].
7. Defences demand precision at the pleading stage. Honest opinion failed because the opinion’s factual foundation was never proved true; reply-to-attack failed because the response was not commensurate with the occasion. Whether proportionality is a separate element of the privilege remains open, but disproportion may evidence malice: at [108] and [131]–[133].
8. Aggravated damages now sit outside the cap and are awarded separately. Post-publication conduct – rejecting a concerns notice, refusing to apologise, doubling down – is fertile ground for aggravation and added $40,000 to the award here: at [149]–[155].
9. The professional obligations run in both directions. For the profession at large, the decision is a reminder that appellate practice is unforgiving of unfocused grounds and points not taken at trial: at [157]–[163]. For Western Australian practitioners specifically, the State’s continued non-adoption of the serious harm element means parallel regimes must be carried in mind in every matter involving online publication: the existence, content and proof of the element may turn entirely on the choice of law analysis under s 11.
10. Conclusion
Latham v Greenwich gives the serious harm element a stable, text-anchored construction and a workable evidentiary framework. The element is neither a nominal hurdle nor a demanding new species of proof: it requires an evaluative judgment, reached on evidence and legitimate inference, that the publication of the matter caused or was likely to cause harm to reputation properly characterised as serious. The Full Court’s rejection of judicial glosses, its location of the element in the publication of the matter rather than in individual imputations, and its endorsement of inferential proof supported by the considerations listed at [64], together supply the analysis that practitioners in the amended jurisdictions – and in the Federal Court – will now apply.
The divergence with Supaphien v Chaiyabarn means the construction question is a candidate for resolution by the High Court, and the Full Court’s observation that the apparently contrary statement in Blake v Fox rests on a slip illustrates how unsettled the comparative landscape remains. In the meantime, the practical message of the decision is direct. Claimants should prove serious harm the way the law has always proved facts that resist direct evidence: by assembling circumstances and inviting inference. Respondents should meet that case with evidence or not at all, and should understand that the conduct of the trial – concessions made, points not taken, apologies withheld – will define both the damages and the available appeal.
For this State’s practitioners the decision is simultaneously foreign and immediate: foreign, because the Defamation Act 2005 (WA) contains no serious harm element; immediate, because the publications that generate modern defamation retainers rarely respect State boundaries. Until Western Australia adopts the 2020 amendments, sound advice on any online publication matter begins with the choice of law and proceeds with both regimes in view.