Continuing detention in Australian constitutional law

The 1996 decision of the High Court in Kable v Director of Public Prosecutions (1996) 189 CLR 51, striking down the validity of the Community Protection Act 1994 (NSW), was a landmark moment for continuing detention regimes in Australia. The provisions of that ad hominem legislation authorised the NSW Supreme Court, on the application of the Director of Public Prosecutions, to order the continuing detention in the custody of the state of Mr Kable who had been convicted of, and was serving a sentence for, the manslaughter of his wife. The order could be made if the Court was satisfied that continuing detention beyond the expiration of his sentence was necessary to protect the public. The reasons for invalidity varied. While the majority judgments were peppered with statements recognising the non-judicial character of a continuing detention power, the thrust of the majority judgments was that the Supreme Court was required by the Act to act in a way that undermined its independence from the other arms of government.

By contrast, eight years later, the Court in Fardon v Attorney-General (Qld) (2004) 223 CLR 575 upheld the Dangerous Prisoners (Sexual Offenders) Act 2003 (Qld), which authorised the Queensland Supreme Court to order the continuing detention of a convicted sex offender if it was necessary to protect the public from the risk of future sex offences. As was the case with the Kable legislation, the Fardon legislation resulted in the deprivation of liberty in the custody of the state beyond a person’s sentence. However, lessons had been learnt from the Kable episode, and the Queensland legislation was drafted to better preserve the judicial process. At least for Gummow J, who had held the Kable legislation invalid, what was important for validity of the Queensland Act was that the Supreme Court’s power to order continuing detention turned on an assessment of the future risk of the person committing offences of the same character as those giving rise to the person’s previous conviction and criminal sentence. To that extent, there remained a connection between the operation of continuing detention provisions and the prior conviction entered in exercise of judicial power. Consequently, if designed correctly, state legislation could authorise continuing detention of a person in the custody of the state beyond the expiration of their criminal sentence.

Continuing detention and the Commonwealth

But that was state legislation. If a federal scheme complied with the Fardon model and preserved independent judicial decision-making, could the Commonwealth Parliament do the same thing? The newly minted Kable limitation set a lower bar for state Parliaments than that which the Commonwealth Parliament must hurdle. While the Kable limitation protects the institutional integrity of state courts, it does not limit them to the exercise of judicial power. By contrast, a court exercising federal jurisdiction can only exercise judicial power (or incidental non-judicial power), and that typically involves the determination of a dispute about pre-existing rights and duties, not the creation of new ones. Additionally, before the High Court could accept Commonwealth continuing detention regimes, it would have to grapple with three iconic statements by Brennan, Deane and Dawson JJ in Chu Kheng Lim v Minister for Immigration (1992) 176 CLR 1 about the exclusive role of courts and the judicial power to protect liberty.

First, their Honours said that ‘putting to one side the exceptional cases …, the involuntary detention of a citizen in custody by the state is penal or punitive in character and, under our system of government, exists only as an incident of the exclusively judicial function of adjudging and punishing criminal guilt’ (at 27).

Secondly, their Honours went further, saying that, leaving aside the exceptional cases, ‘the citizens of this country enjoy, at least in times of peace …, a constitutional immunity from being imprisoned by Commonwealth authority except pursuant to an order by a court in the exercise of the judicial power of the Commonwealth’ (at 28-9).

Thirdly, when the character of a power of detention was being determined, the enquiry was not just directed to identifying a non-punitive purpose: the Lim joint judgment also asked whether the law is ‘reasonably capable of being seen as necessary’ for the non-punitive purpose (at 33). If not, then it would be seen as punitive in character.

These statements appeared to straddle two approaches for determining whether detention would be exclusively judicial. On the one hand, there was a general rule that detention required an exercise of judicial power through a criminal trial unless exceptional circumstances could be shown. On the other hand, there was resort to the purpose of detention – whether it was punitive or non-punitive – to characterise the power. The two approaches could be reconciled if the exceptional categories were identified by reference to non-punitive purposes, although questions would inevitably arise as to how to contain the exceptions, in particular, whether certain non-punitive purposes might be considered illegitimate. It was also unclear whether a power of continuing detention for non-punitive purposes (an indicator of the non-judicial character of immigration detention in the Lim context) could be given to courts to be exercised according to a judicial process other than a criminal trial. However, because the context of Lim involved an established exception to the general rule – detention by the executive for the non-punitive purpose of deportation – these subtleties were not examined further.

In speculating about a Commonwealth continuing detention regime, McHugh J and Gummow J in Fardon engaged with these Lim statements in different ways. Justice McHugh, favouring the view that the character of detention was to be determined by its punitive/non-punitive purpose, took the view that the first Lim statement had been overstated and that the second Lim statement of a constitutional immunity could not stand (at 24). For his Honour, whether detention required an exercise of judicial power according to the criminal process turned on whether the legislative purpose was punitive, and there was no reason in principle why power to order continuing detention for protective, non-punitive purposes could not be given to a court by the Commonwealth Parliament. While continuing detention lacked the classic judicial power characteristic of a determination of a dispute about pre-existing rights, McHugh J was of the view that a continuing detention power was of the ‘same jurisprudential character’ as orders traditionally made by courts determining a particular fact or status, as in cases of matrimonial causes, bankruptcy, probate and the winding up of companies (at 597). However, unless the Lim approach of a general rule subject to exceptions were to be abandoned entirely in favour of an approach that looked exclusively to purpose to determine the character of detention, then it was not clear how McHugh J’s ‘same jurisprudential character’ approach could be gelled with Lim. While it might be readily seen as determining a fact or status, why was it judicial power rather than non-judicial power?

By contrast, Gummow J considered the distinction between punitive and non-punitive purposes to be unhelpful, and that the first Lim statement should instead turn directly on the deprivation of liberty: exceptional cases aside, ‘the involuntary detention of a citizen in custody by the State is permissible only as a consequential step in the adjudication of criminal guilt of that citizen for past acts’ (at 611). While state Parliaments could, if the scheme were properly designed, confer a power of continuing detention on state courts, the Commonwealth Parliament could not. Continuing detention could not be seen as analogous to the exceptional categories, like the treatment of the mentally ill, to which the Lim limitation was subject. By eschewing the punitive/non-punitive purpose framework, Gummow J was able to conclude that continuing detention without a criminal trial was beyond power: it could not be non-judicial; nor could it be exercised by courts if it did not follow from the adjudgment and punishment of criminal guilt.

A twist in the story

When Mr Kable returned to the NSW Supreme Court seeking compensation for his unlawful detention, it seemed that McHugh J’s conclusion, that continuing detention is judicial in character when given to a court, had prevailed in the High Court. The High Court in New South Wales v Kable (2013) 252 CLR 118 took the view that Kable was not entitled to compensation because at all times he remained in detention pursuant to the order of a superior court which remained valid until set aside. Although the Court did not engage directly with the question of whether the Commonwealth Parliament could enact continuing detention laws, and without discussion of the competing views of McHugh J and Gummow J in Fardon, their Honours in the 2013 Kable case referred to the continuing detention order against Kable as having been made in the exercise of judicial power. However, because the question before the Court was not focused on the character of the power exercised by the Supreme Court, the conclusion on the character of power was unclear and undeveloped.

Div 105A of the Criminal Code (Cth) and Benbrika

Against this background, the Commonwealth Parliament enacted Division 105A of the Criminal Code (Cth) empowering the Supreme Court of a State or Territory, on application by the Minister for Home Affairs, to order that a person convicted of certain terrorist offences, including a serious Part 5 terrorist offence (a Part 5 terrorist offence where the maximum penalty is seven years or more), be detained in prison following the expiration of their sentence. It was drafted to be Fardon-proof; to avoid the potential pitfalls into which the Kable legislation fell. Yet, as Commonwealth legislation, the constitutional standard to be met is whether the function conferred is judicial in character – not simply whether the judicial process remains sufficiently intact. That would require engagement with the core characteristic of judicial power – the determination of a dispute about pre-existing rights and duties – and the Lim joint judgment’s three statements.

In 2008, the respondent, Abdul Nacer Banbrika, had been convicted of two terrorist offences. He was sentenced to a term of imprisonment of 15 years and his sentence expired on 5 November 2020. On 4 September 2020, the Minister for Home Affairs sought a continuing detention order from the Victorian Supreme Court. A continuing detention order was made under Division 105A for a period of three years. The object of Division 105A ‘is to ensure the safety and protection of the community by providing for the continuing detention of terrorist offenders who pose an unacceptable risk of committing serious Part 5.3 offences if released into the community’ (s 105A.1). A continuing detention order can only be made by a Supreme Court: (i) against a person who has been convicted of a terrorist offence of a particular kind (including a serious Part 5.3 offence); (ii) where the person has been in continuous custody since the conviction; (iii) where the person will be at least 18 years old at the end of the sentence; (iv) where the Court ‘is satisfied to a high degree of probability, on the basis of admissible evidence, that the offender poses an unacceptable risk of committing a serious Part 5.3 offence if the offender is released into the community’; and (v) where the Court ‘is satisfied that there is no other less restrictive measure that would be effective in preventing the unacceptable risk’ (ss 105A.3 and 105A.7(1)). The period of continuing detention can be up to three years, but no more than the Court is ‘satisfied is reasonably necessary to prevent the unacceptable risk’ (s 105A.7(5)). Successive orders are permitted. Section 105A.8 sets out a range of matters to be taken into account, including the safety and protection of the community, reports and assessments in relation to the offender, the offender’s conviction history, the views of the sentencing court and any other information as to the risk of the offender committing a serious Part 5.3 offence. The orders are to be reviewed annually, and must be revoked unless the Court is satisfied that the offender continues to pose an unacceptable risk and that there are no less restrictive means of addressing the risk.

Continuing detention can be judicial in the hands of a court

In Minister for Home Affairs v Benbrika, decided in February 2021, six judges held that the power to order continuing detention can be conferred by the Commonwealth Parliament on courts (Gordon J did not express a concluded view). Justice Gummow’s reformulation of Lim from Fardon did not find favour. Nor was it accepted that a continuing detention power had to protect against the commission of an offence of the same character as that for which the offender was originally punished. Thus, Gummow J’s reason for upholding the Fardon legislation against a Kable challenge was not transplanted into the Lim principles.

However, nor were their Honours content to rely upon McHugh J’s explanation in Fardon for why the power of continuing detention can be judicial. For five judges, continuing detention by a court was held to fall within an exception to the Lim general rule. For Kiefel CJ, Bell, Keane and Steward JJ, continuing detention had a protective purpose and was sufficiently analogous to court orders for the detention of the mentally ill (at [36]) (again, in this respect, Gummow J’s views did not find favour).

Thus, while McHugh J’s statement about the jurisprudential similarity to other powers determining a fact or status was referred to in the joint judgment with approval (at [14]), the rationale for that jurisprudential similarity was revealed through analogy to an exceptional category of power, permitting the creation of rights and duties, protective of the community and historically exercised by courts. Justice Gageler, who dissented, accepted that a continuing detention power could be seen as a double function provision which ‘creat[ed] a liability to be detained through an act of adjudication …’ (at [76]). It could be exercised by a court without a criminal trial because it fell within an exceptional category. Thus, for these five judges, the power to order continuing detention can be judicial in the hands of a court even though the process does not involve a criminal trial. It is reconciled within the first Lim statement as an exception to the general rule that detention is penal and to be effected only through the adjudgment and punishment of criminal guilt. Justice Gordon, in dissent, reserved judgment on whether a new exceptional category should be recognised.

By contrast, Edelman J sought to reconceptualise the Lim notion of punishment: continuing detention is ‘a form of protective punishment’ (at [182]) which, subject to possible exceptions, is exclusively judicial, but which could be exercised according to a judicial process that did not involve a criminal trial.

Diverging views on the scope of the judicial power of continuing detention

Despite the general agreement of six judges that the Commonwealth Parliament can confer continuing detention powers on courts for the purpose of protecting the community, the Court was divided on its approach to the leeway permitted to the legislature in designing a continuing detention scheme.

Part 5.3 of the Criminal Code creates a wide range of terrorism offences, with varying degrees of connection to a terrorist act. As Gordon J said, ‘[t]he nature and extent of the harm that may be caused to persons or property by commission of a serious Part 5.3 offence will vary widely’ (at [16]); the offences ‘are not restricted to offences or conduct having an immediate harm to persons or property (at [169]). As Gageler J put it, most of the offences in Part 5.3 are ‘prophylactic’ (at [55]).

The majority

For Kiefel CJ, Bell, Keane and Steward JJ, the breadth of the offences to be protected against did not lead to invalidity. Their Honours said (at [46]):

It is difficult to envisage any circumstances in which a continuing detention order would be made to prevent the risk of the commission of a serious Pt 5 offence where the offence is of a kind that could not be seen to pose a real threat of harm to the community. Even where the apprehended serious Pt 5.3 offence does not involve as an element the inflicting, or having as an immediate purpose the actual inflicting, of personal injury on a person or persons, the advancement of terrorist ideology can readily be seen to create a milieu which fosters the prospect that personal injury will be suffered by innocent members of the community. A law directed against the implementation of such an ideology (even by preparatory acts) does not lack the character of a law for the protection of the community from harm simply because the law does not include the immediate likelihood or purpose of inflicting personal injury as an element of the offence.

It was ‘important’, their Honours said, that ‘the restriction upon liberty involved in the making of a continuing detention order is dependent upon the risk of an offence being “unacceptable” to the judge in light of the facts as they appear at the time he or she is asked to make the order’. Furthermore, the power to make an order, such as a control order, ‘less intrusive on personal liberty … serves to ensure that continuing detention orders are made to secure the protection of the community from unacceptable risks of actual harm’ (at [46]). ‘Correctly understood’, their Honours concluded, ‘a continuing detention order could not properly be made by a Court in the exercise of the discretion … in a case where the only risk of offending identified by the authorities did not carry a threat of harm to members of the community that was sufficiently serious in the assessment of the Court as to make the risk of the commission of the offence “unacceptable” to that Court’. The provisions remained ‘rightly characterised as directed to ensuring the safety and protection of the community from the risk of harm posed by the threat of terrorism’ (at [47]).

Three points may be made about this analysis. First, while not explained as such, this analysis appears to show that the joint judgment considered whether the detention scheme is ‘appropriately tailored’ for the protection of harm (an expression used by their Honours earlier at [32] when rejecting Gummow J’s approach from Fardon). However, consistently with what was said by Kiefel CJ, Bell, Keane and Edelman JJ in Falzon v Minister for Immigration (2018) 262 CLR 333, it seems that the reason why that analysis was undertaken was to determine whether the ‘true purpose’ (Falzon at 344) of the law is to protect against a risk to the community. Under such an approach, the requisite degree of fit between the means and end looks to be a relaxed one, perhaps resembling the analysis in other areas of constitutional law that requires only a rational connection.

Secondly, in an application of that relaxed means/end test, it did not matter that the court’s detention power is couched in terms of preventing the risk of the commission of a serious Part 5 offence, rather than being directed expressly to the prevention of a risk of harm. Because the risk of harm to the community can be considered when determining whether the risk of an offence being committed is ‘unacceptable’, and because a court is to apply less restrictive measures instead of continuing detention orders if that would be effective to prevent that unacceptable risk from arising, then the regime is appropriately tailored to the purpose of protecting the community.

Thirdly, nonetheless, it remains valid to cast the net widely to prevent conduct that does not inherently lead to harm. Such conduct, when viewed in the context of all the facts, might carry a risk of harm depending on the circumstances. Deprivation of liberty may be inflicted as a preventive measure, not only when there is a risk of future conduct that carries an inherent risk of harm, but also when conduct, lacking that inherent risk, otherwise fosters the creation of such a risk.

It was a disagreement with the second and third points which explained the dissenting judgments of, respectively, Gordon J and Gageler J.

Dissenting judgments

For Gordon J, if the breadth of the offences to be prevented, which covered conduct with no immediate harm, was explicable on the basis that the commission of any of those offences carried with it ‘a high degree of probability of serious harm’, then that should have been the condition for the making of the order (at [169]). The use of the word ‘unacceptable’ could not cure that deficiency: a Supreme Court’s satisfaction of the unacceptability of risk was directed to the commission of an offence – not to harm that might occur. ‘The concern’, her Honour said, ‘is not of harm, but of the offender committing an offence regardless of the consequence of that offending for the community’ (at [170]). Additionally, the assessment of other less restrictive measure was directed to the former risk – not the latter. ‘It is not to the point’, her Honour said

that, in some cases, preparatory or anticipatory acts might in some way or other advance an ideology that increases the possibility that harm to persons or property will be caused at some time in the future. The relevant inquiry is not how a Supreme Court might reason when applying the criteria in s 105A.7(1) in a given case but the proper construction of Div 105A.

The short answer was that the substantive criteria for the making of an order ‘turn on an unacceptable risk of committing a serious Part 5.3 offence, not what consequences the commission of the offence might entail’ (at [176]). Consequently, the criteria for making the order were ‘too broad’ (at [175]) and thus the detention regime ‘was not sufficiently tailored to the stated purpose of Div 105A to be an exercise of judicial power’ (at [177]): Div 105A ‘goes further than necessary to achieve the objective’ (at [177]).

Justice Gageler’s dissent similarly considered that continuing detention had to be justified by ‘close scrutiny’ (at [99] per Gageler J) and that, consequently, there needs to be a ‘demanding constitutional justification’ (at [73]). Furthermore, the detention must be, as set out in Lim, ‘reasonably capable of being seen as necessary for a legitimate non-punitive objective’ (at [78]). His Honour accepted that ‘[p]revention of harm is a non-punitive objective, at least where the harm is grave and specific’ (at [79]). However, the ‘objective merely of preventing commission of a criminal offence cannot be legitimate’ (at [85]); otherwise, it would undermine the paradigm position that the determination of criminal guilt for an offence must be through a criminal trial. The impugned provisions were directed to preventing the commission of serious Part 5.3 offences. While, in some circumstances, the prevention of the commission of certain offences might serve as a proxy for a legitimate purpose of protecting the community from serious harm (his Honour explained Fardon on this basis), the problem was that the category of serious Part 5.3 offences was too broad: it covered conduct which was not ‘inherently harmful’ (at [97]). The ‘prophylactic approach’ to criminalising conduct under Part 5.3 meant that the offences covered ‘conduct many steps removed from doing or supporting or facilitating any terrorist act’ (at [93]). Further, while the parliamentary record revealed a purpose of preventing a terrorist act (considered by his Honour to be a legitimate objective), the means chosen (ie, the wide coverage of offences to be prevented) lacked a ‘close correspondence’ to that legitimate objective (at [95]): that is, the detention regime was overly broad. In conclusion, Gageler J was prepared to support the validity of Div 105A to the extent that the serious Part 5 offences to be protected against involved ‘doing or supporting or facilitating a terrorist act’ (at [64]).

It is evident that the dissenting judges applied a closer form of scrutiny. For Gordon J, the detention regime could have been better designed to focus on the risk of harm to the community. The requisite degree of connection could not be achieved by straining the language of s 105A.7(1) to allow the same risk assessment through the word ‘unacceptable’. For Gageler J, the protective purpose had to be legitimate (preventing crime is not enough) and there needed to be a close fit between the legislative means and the legitimate end.

The judgment of Edelman J

The other member of the majority, Edelman J, took a different approach. Protective punishment within Ch III is directed to the prevention of crime (at [228]). Consistency with Ch III required ‘a form of structured proportionality analysis’ (at [226]), which would not be satisfied if (i) ‘the purpose of the protective punishment could easily be met to the same extent by reasonable alternatives, such as less restrictive control orders, which could achieve the statutory purpose without the extreme constraint upon liberty of detention’ or (ii) ‘the purpose for the protective punishment, assessed primarily by reference to the importance placed upon that purpose by Parliament, is so slight or trivial that it cannot justify detention of an individual’. The latter circumstance would only arise in ‘extreme cases’ as it is not the job of the courts to refuse to implement a law ‘on the grounds of an objection to legislative policy’ (at [226]).


There is much to unpack from Benbrika. However, five short points will be noted here.

First, it is clear that the difference of opinion between McHugh J and Gummow J in Fardon has now been addressed. While Gummow J’s approach has not found favour, McHugh J’s challenge to the first Lim statement has not prevailed. The Court has explained why continuing detention is judicial in character, despite lacking the core characteristic of determining a dispute about pre-existing rights and duties. It has also rationalised continuing detention within the first Lim statement as an exception to the general rule, thereby explaining how it can be exercised by courts even in the absence of a criminal trial.

Secondly, it is evident that the Court is divided on the scope of Commonwealth continuing detention regimes. The joint judgment of Kiefel CJ, Bell, Keane and Steward JJ leaves considerable latitude for Parliament to identify a class of offences which, when viewed by a court in the context of particular facts presenting a risk of harm to the community, could be made the subject of a continuing detention regime. That is the case even though a subset of those offences criminalise conduct that is not inherently harmful. Justice Gageler disagrees. It might be said that, on the one hand, as recognised by Gageler J, his Honour’s view gives rise to a necessary exercise in line drawing. On the other hand, an approach which allows the Parliament too much leeway runs the risk of subverting the very foundation of criminal justice as involving a trial of a criminal offence for past conduct. Whether Gordon J would be prepared to accept the majority position is unclear. It seems most likely that, like Gageler J, her Honour would require the link between conduct and risk to appear on the face of the legislation rather than being left to the evaluation of the court ordering detention. Of course, because of the particular approach taken by Edelman J, his Honour would seem to endorse the prevention of any crime as a legitimate end, provided the threat posed is not trivial.

Thirdly, as to the requisite degree of fit between the means of detention and the non-punitive purpose, a majority of the Court, first in Falzon, and now in Benbrika, must be taken to understand the third Lim statement as permitting a relaxed connection between the means and end, something close to a rationality standard used in other constitutional contexts. The connection between the means and end is only scrutinised to double check that there truly is a non-punitive purpose. By contrast, Gageler J and Gordon J, in dissent, see the rigorous testing of the means and end as an important way of ensuring that the cornerstone of criminal justice protected by Ch III is not subverted.

Fourthly, undoubtedly informing these conclusions is the value placed on the role of Ch III in protecting liberty. That value, at the core of the joint judgment in Lim, was the driving force behind Gummow J’s attempt to rework the Lim principles. While Gageler J and Gordon J declined to accept Gummow J’s approach, it is clear that their Honours accept that (i) the protection of liberty lies at the heart of Ch III (at [68]-[74], [98]-[99] (Gageler J); [130]-[140] (Gordon J)), and (ii) consequently, departures from the general rule, that detention follows a criminal trial, must be narrowly conceded in terms of identifying the degree of fit between the means of detention and the legitimate purpose. For Gageler J, it also means that non-punitive purposes must be carefully scrutinised for legitimacy.

By contrast, the majority judges did not agree. Echoing McHugh J in Fardon, Edelman J is the clearest in this respect: ‘there has never been any independent constitutional principle of individual liberty that denies to the State the power to implement a policy choice that deprivation of liberty is required for an orderly society’ (at [216]). It appears that Kiefel CJ, Bell, Keane and Steward JJ take a similar view, saying that ‘Chapter III courts serve as the bulwark of liberty’, but it would seem only ‘by virtue of the qualities of independence and impartiality that are secured by the separation of the judicial function from the other functions of government’ (at [20]). Thus, liberty is protected as a consequence of an independent judiciary, rather than an independent judiciary protecting liberty as a freestanding value protected by Ch III. This is consistent with the position accepted by Kiefel CJ, Bell, Keane and Edelman JJ in Falzon that Lim did not ‘establish that there is a constitutionally guaranteed freedom from executive detention’ (at 343). There appears little left of the second Lim statement about a ‘constitutional immunity’.

Finally, it is hard to resist the conclusion that the consequence of the decision, for designing government decision-making, was an important consideration. ‘Terrorism’, it was said in the joint judgment, ‘poses a singular threat to civil society’ (at [36]). For Kiefel CJ, Bell, Keane and Steward JJ, it was unattractive to conclude that there was a lacuna of power to order continuing detention (which would have been the case if Gummow J’s view had been adopted), and ‘distinctly unattractive’ (at [30]) to conclude that the executive, but not the courts, could exercise such a power (which would have resulted if an exception to Lim could not be crafted). In leaving the door open for courts to create rights of detention through continuing detention orders, Gageler J was also ‘sensitive to the reality’ (at [77]) that it is a good thing for courts to protect liberty.

James Stellios is a Professor at the ANU Law School

Suggested citation: James Stellios, ‘Minister for Home Affairs v Benbrika [2021] HCA 4’ on AUSPUBLAW (14 April 2021) <>