05 July 2024

GenAI Patents

The WIPO Patent Landscape Report on Generative Artificial Intelligence comments 

 The release of OpenAI’s ChatGPT chatbot in November 2022 has greatly increased public enthusiasm for generative AI (GenAI). It has been described by many, including Nvidia CEO JenHsun Huang, as an “iPhone moment” for GenAI. This is because the OpenAI platform has made it easier for all users to access advanced GenAI programs, particularly large language models (LLMs). These models have reached new levels of performance, demonstrating the potential for various real-world applications, triggering a wave of research and development, and large corporate investments in GenAI. 

This WIPO Patent Landscape Report provides observations on patenting activity and scientific publications in the field of GenAI and builds on the 2019 WIPO Technology Trends publication on Artificial Intelligence. It aims to shed light on the current technology development, its changing dynamics and the applications in which GenAI technologies are expected to be used. It also identifies key research countries, companies and organizations. 

GenAI patent families and scientific publications have increased significantly since 2017 

The rise of GenAI over the past few years has been driven primarily by three factors: more powerful computers, the availability of large datasets as a source of training data, and improved AI/machine learning algorithms. Developments such as the transformer architecture in LLMs have significantly advanced GenAI. This has made it possible to develop complex applications in many different fields. 

The technological advances in GenAI are reflected by the sharp increase in patenting activity. Over the past 10 years, the number of patent families in GenAI has grown from just only 733 in 2014 to more than 14,000 in 2023. Since the introduction of the transformer in 2017, the deep neural network architecture behind the Large Language Models that have become synonymous with GenAI, the number of GenAI patents has increased by over 800%. The number of scientific publications has increased even more over the same period, from just 116 in 2014 to more than 34,000 in 2023. Over 25% of all GenAI patents and over 45% of all GenAI scientific papers were published in 2023 alone. 

Which are the top organizations with the most patents in GenAI?

1. Tencent 

2. Ping An Insurance Group 

3. Baidu 

4. Chinese Academy of Sciences 

5. IBM

 Tencent, Ping An Insurance Group and Baidu own the most GenAI patents. Tencent plans to add GenAI capabilities to its products such as WeChat to improve the user experience. Ping An focuses on GenAI models for underwriting and risk assessment. Baidu was one of the early players in GenAI and recently unveiled its latest LLM-based AI chatbot, ERNIE 4.0. The Chinese Academy of Sciences (fourth) is the only research organization in the top 10 ranking. Alibaba (sixth) and Bytedance (ninth) are other Chinese companies in the top 10. 

IBM (fifth), Alphabet/Google (eighth) and Microsoft (10th) are the top US companies in terms of GenAI patents. IBM has developed a GenAI platform, watsonx, which enables companies to deploy and customize LLMs with a focus on data security and compliance. Alphabet/Google's AI division DeepMind recently released its latest LLM model, Gemini, which is gradually being integrated into Alphabet/Google's products and services. Microsoft is another key player in GenAI and an investor in OpenAI. OpenAI itself has only recently filed its first GenAI patents. Rounding out the top 10 is electronics conglomerate Samsung Electronics (seventh) from the Republic of Korea.

Which institutions published the most scientific publications on GenAI? 

The Chinese Academy of Sciences is clearly in the lead in terms of scientific publications with more than 1,100 publications since 2010. Tsinghua University and Stanford University follow in second and third place with more than 600 publications each. Alphabet/Google (fourth) is the only company in the top 20 (556 scientific publications). 

However, when measuring the impact of scientific publications by the number of citations, companies dominate. Alphabet/Google is the leading institution by a wide margin, and seven other companies are present in the top 20. The case of OpenAI is also noteworthy. In our GenAI corpus of scientific publications, the company has published only 48 articles (325th institution in terms of number of publications), but these publications have received a total of 11,816 citations from other scientific publications (13th overall). 

Where are the most GenAI technologies invented?

1. China 

2. United States 

3. Republic of Korea 

4. Japan 

5. India 

6. United Kingdom 

7. Germany

Inventors based in China were responsible for more than 38,000 patent families between 2014 and 2023, based on the inventor addresses published on patents. Since 2017, China has published more patents in this field each year than all other countries combined. 

With around 6,300 patent families between 2014 and 2023, the US is the second most important research location for GenAI patenting. The Asian countries Republic of Korea, Japan and India are other key research locations for GenAI, all ranking in the top 5 countries worldwide (third, fourth and fifth respectively). The United Kingdom is the leading European location (sixth globally), with 714 patents published in the same period. However, Germany is close behind (708 patent families) and has published more GenAI patents than the UK in recent years. These top inventor locations account for the majority (94%) of global patenting activity related to GenAI. 

Which GenAI model has the most patents? 

In recent years, a number of GenAI programs, or models, have been developed. Among the most important GenAI models are:

1. generative adversarial networks (GANs) 

2. variational autoencoders (VAEs) 

3. decoder-based large language models (LLMs) 

However, not all GenAI patents can be assigned to these three specific core models based on available information from patent abstracts, claims or titles. 

Among these GenAI models, most patents belong to GANs. Between 2014 and 2023, there were 9,700 patent families of this model type, with 2,400 patent families published in 2023 alone. VAEs and LLMs are the second and third largest models in terms of patents, with around 1,800 and 1,300 new patent families respectively between 2014 and 2023. 

In terms of patent growth, GAN patents show the strongest increase over the past decade. However, this has slowed down recently. In contrast, diffusion models and LLMs show much higher growth rates over the last three years, with the number of patent families for diffusion models increasing from 18 in 2020 to 441 in 2023 and for LLMs increasing from 53 in 2020 to 881 in 2023. The GenAI boom caused by modern chatbots such as ChatGPT has clearly increased research interest in LLMs. 

What are the main types of data used in GenAI patents? 

The main GenAI data types include: – Image – Video – Speech – Sound – Music 

Among the different GenAI modes, or the type of data input and output, most patents belong to the image/video category. Image/video data is particularly important for GANs. Patents involving the processing of text and speech/sound/music are key data types for LLMs. The remaining modes: 3D image models, chemical molecules/genes/proteins and code/software have far fewer patents so far. As with patents related to GenAI core models, some patents cannot be clearly assigned to a specific data type. In addition, some patents are assigned to more than one mode because certain GenAI models, such as multimodal large language models (MLLMs), overcome the limitation of using only one type of data input or output. 

Top application areas of GenAI patents 

The key application areas for GenAI patents include: 1. Software 2. Life sciences 3. Document management and publishing 4. Business solutions 5. Industry and manufacturing 6. Transportation 7. Security 8. Telecommunications 

GenAI is bound to have a significant impact on many industries as it finds its way into products, services and processes, becoming a technological enabler for content creation and productivity improvement. For example, there are many GenAI patents in life sciences (5,346 patent families between 2014 and 2023) and document management and publishing (4,976). Other notable applications with GenAI patents ranging from around 2,000 to around 5,000 over the same period are business solutions, industry and manufacturing, transportation, security and telecommunications.  

In the life sciences sector, GenAI can expedite drug development by screening and designing molecules for new drug formulations and personalized medicine. In document management and publishing, GenAI can automate tasks, save time and money, and create tailored marketing materials. In business solutions, GenAI can be used for customer service chatbots, retail assistance systems, and employee knowledge retrieval. In industry and manufacturing, GenAI enables new features like product design optimization and digital twin programming. In transportation, GenAI plays a crucial role in autonomous driving and public transportation optimization. 

However, many patent families (around 29,900 patent families between 2014 and 2023) cannot be assigned to a specific application based on the patent abstract, claims or title. These patents are instead included in the category software/other applications.

02 July 2024

Smoke Screening

“Keep it a secret”: leaked documents suggest Philip Morris International, and its Japanese affiliate, continue to exploit science for profit' by Sophie Braznell, Louis Laurence, Iona Fitzpatrick and Anna B Gilmore in (2024) Nicotine and Tobacco Research comments 

The tobacco industry has a long history of manipulating science to conceal the harms of its products. As part of its proclaimed transformation, the world’s largest tobacco company, Philip Morris International (PMI), states it conducts “transparent science”. This paper uses recently leaked documents from PMI and its Japanese affiliate, Philip Morris Japan (PMJ), to examine its contemporary scientific practices. ... 

23 documents dating 2012 through 2020 available from Truth Tobacco Industry Documents Library were examined using Forster's hermeneutic approach to analysing corporate documentation. Thematic analysis using the Science for Profit Model was conducted to assess whether PMI/PMJ employed known corporate strategies to influence science in their interests. 

... PMJ contracted a third-party external research organisation, CMIC, to covertly fund a study on smoking cessation conducted by Kyoto University academics. No public record of PMJ’s funding or involvement in this study was found. PMJ paid life sciences consultancy, FTI-Innovations, ¥3,000,000 (approx. £20,000) a month between 2014 and 2019 to undertake extensive science-adjacent work, including building relationships with key scientific opinion leaders and using academic events to promote PMI’s science, products and messaging. FTI-Innovation’s work was hidden internally and externally. These activities resemble known strategies to influence the conduct, publication and reach of science, and conceal scientific activities. ... 

The documents reveal PMI/PMJ’s recent activities mirror past practices to manipulate science, undermining PMI’s proclaimed transformation. Tobacco industry scientific practices remain a threat to public health, highlighting the urgent need for reform to protect science from the tobacco industry’s vested interests. ... 

Japan is a key market for PMI, being a launch market for IQOS and having the highest heated tobacco product use globally. Our findings, in conjunction with other recent evidence, challenge PMI’s assertion that it is a source of credible science and cast doubt on the quality and ethical defensibility of its research, especially its studies conducted in Japan. This, in turn, brings into question the true public health impacts of its products. There is urgent need to reform the way tobacco-related science is funded and conducted. Implementation of models through which research can be funded using the industry’s profits while minimising its influence should be explored. ... 

Science is essential to understanding and improving public health. Unfortunately, science is also used as a tool by corporate actors across diverse industries to conceal or create doubt about the harms of their products or manufacturing activities; to position their products as solutions to complex problems; and to legitimise their role in both science and policy. Corporate misuse of science is detrimental to public health as it delays and weakens policies, prevents litigation to protect consumers, and maximises use of potentially damaging products. These mechanisms ultimately serve to maximise corporate profits rather than primarily improve public health. Tobacco companies have a particularly well-documented history of scientific misconduct because of the release of internal industry documents which revealed they repeatedly prioritised their bottom line over the health of billions of people. 

Philip Morris International (PMI), the largest transnational tobacco company in the world, played a prominent role in the tobacco industry’s history of scientific misconduct and manipulation. In 2016, under threat from declining cigarette sales, PMI announced that it would be undergoing a “smoke- free” transformation, with the aim of replacing its cigarettes with its newer nicotine and tobacco products. Its flagship newer product brand is a heated tobacco product (HTP), IQOS, which it claims is less harmful than cigarettes. 

Scientific research is fundamental to PMI’s claimed transformation and substantiating the harm reduction claims it makes about its newer products. PMI promotes its role in science to both the public and policymakers, proclaiming it conducts “transparent science”. Yet recent evidence contradicts this claim and raises questions about whether PMI has truly transformed. For example, the Foundation for a Smoke-Free World (FSFW) claims to be independent but is solely funded by PMI and publishes PMI-favourable research, mirroring the long-standing tobacco industry practice of using scientific third parties to promote products and corporate messaging. Moreover, journalist investigations and academic reviews of PMI’s science have raised serious concerns over the quality and ethical standing of PMI’s clinical research. 

In this context, a small sample of recently leaked documents relating to the scientific activities of PMI and its Japanese affiliate, Philip Morris Japan (PMJ), provide limited but unique insight into PMI’s contemporary scientific practices. Japan is of particular interest because it has the highest prevalence of HTP use of any country worldwide. PMI therefore often uses Japan as the location for its studies on HTPs, including those submitted to regulatory bodies in the US and EU, and publicly promotes Japan as “an example of successful harm reduction to other countries”. The aims of this study were to examine these new documents in order to gain insight into PMI’s contemporary research practices and use of science; assess whether PMI continues to employ known strategies to influence science; and discuss the implications of its practices for wider tobacco research.

Secrecy

The Independent National Security Legislation Monitor report on Secrecy Offences – Review of Part 5.6 of the Criminal Code Act 1995 comments 

 There is some information that, in the wrong hands, could genuinely harm Australia’s national interests. Secrecy offences that criminalise unauthorised dealing with and communication of this type of information have a role to play in deterrence and punishment. This review considers whether the offences in Part 5.6 of the Criminal Code operate in a way that is effective, necessary, proportionate and consistent with Australia’s international obligations; and whether the offences contain sufficient safeguards. 

There are problems with the current offences, and these need to be addressed. For example, there is significant uncertainty as well as conflict with rule of law principles because of the way a policy framework is used to define key elements of some offences. There are also problems with proportionality in penalising non-officials, particularly in relation to receiving or otherwise dealing with information. The recommendations of this review will ensure that information which is or is likely to be harmful to critical national interests remains protected but in a way that is clearer and more consistent with the rule of law and concepts of necessity and proportionality; and with Australia’s international obligations. 

 This report begins with an overview of the context in which Part 5.6 of the Criminal Code operates and key definitions (Chapter 1), the current threat environment (Chapter 2) and international obligations (Chapter 3). Chapter 4 then considers the main deemed harm offence, finding it uncertain, inconsistent with key rule of law principles and unreasonably broad. Recommendations 1−5 are made to address these concerns, the most important being the removal of reliance on security classification markings as an element of an offence. Harm-based and related offences for officials are analysed in Chapters 5−7. Recommendations 6, 7, 9 and 10 are directed at improving these, including by clarifying definitions and ensuring a cascading approach to penalties. Recommendation 11 sets out principles for a replacement general offence for officials. A key recommendation concerning offences for non-officials is removing ‘dealing with’ offences (Recommendation 8) along with refining the circumstances in which a disclosure offence occurs (Recommendation 12). While there is some scope to recast the defence for journalists as an exception, it is not recommended that a general public interest test or element be introduced (Recommendation 13). Safeguards in both policy and the legislation are discussed in Chapter 10, with some enhancements suggested (Recommendations 14 and 15). 

In conducting this review, I consulted with a range of stakeholders, published a detailed Issues Paper, held individual and roundtable meetings, received 22 written submissions, held 2 days of public hearings, received 4 supplementary submissions and responses to questions on notice from 6 agencies. I was greatly assisted by considered submissions from non- government organisations, government agencies and academics as well as by counsel and INSLM staff. I thank all of those who have been involved in the review. 

Deemed harm offences for Commonwealth officials 

Deemed harm offences do not require proof of any actual harm or risk of harm. They should only be used for information that is always, or almost always, going to be harmful to a critical national interest if disclosed. The current deemed harm offence in s 122.1 of the Criminal Code applies to three broad categories of information: any information that has been classified as secret or top secret; any information obtained or made by or for one of Australia’s 6 main intelligence agencies or any foreign intelligence agency in connection with its functions; and any information about the operations, capabilities or technologies of, or methods or sources used by, any domestic or foreign law enforcement agency. 

Security classified information 

Relying on classification markings to guide how officials use, store and disseminate information is a longstanding and reasonable policy. However, seeking to incorporate a policy directly into the criminal law raises many legal issues. 

Information is ‘security classified’ if it has a security classification applied ‘in accordance with’ a policy framework. There is real uncertainty about when a classification is applied ‘in accordance with’ the policy framework. Not all information marked ‘secret’ or ‘top secret’ will meet the legal test. It is likely that evidence will need to be led in a prosecution to establish that disclosure of the information would be expected to cause serious or grave damage to specified interests. The current policy framework lacks the precision expected for criminal law and parts appear inconsistent with the requirements in the Criminal Code. Training, record-keeping and review procedures are inadequate when it comes to making decisions that have such significant criminal law implications. These things may not be necessary for a policy that has only administrative consequences, but should be in place for decision-making that has such serious criminal law consequences. 

Even if more certainty could be injected into classification decision making there are fundamental rule of law issues with having, as a core element of an offence, decisions made under a policy framework. This is because the executive can change policies at any time. It is also problematic that parts of the relevant policy framework are not publicly available. Furthermore, reliance on a policy document to frame a criminal offence may not be compliant with Australia’s international obligation that limits to the freedom of expression can only be imposed by law. 

Removing classification markings as a physical element of the offences does not mean that they have no role to play whatsoever. A classification marking combined with evidence of an official’s training and experience in what those markings indicate, would go a long way towards establishing, as matter of fact, the required fault element of recklessness. 

RECOMMENDATION 1: The offences in Part 5.6 should not rely on information being classified under a policy framework as an element of the offence. 

Information about intelligence agencies 

Disclosure of information about the core intelligence operations, capabilities, technologies, methods and sources of Australia’s 6 main intelligence agencies will almost always be harmful to the national interest. But it does not follow that all information to do with those agencies will always be harmful. This is particularly so for agencies whose functions have expanded beyond traditional intelligence functions and now include assisting other bodies – such as the Department of Home Affairs – with their functions; undertaking broad cybersecurity functions; or some commercial mapping. The same is true for routine administrative and corporate information. Disclosure of information about non-intelligence functions and administrative activities can result in harm in some circumstances but not with the degree of certainty that justifies a deemed harm offence. Where disclosures about these other activities cause harm or are likely to cause harm to security, defence or international relations, they will be covered by the harm-based offence in s 122.2 (Recommendation 6). 

RECOMMENDATION 2: The deemed harm offences in s 122.1 should not apply to all information connected to an intelligence agency’s functions. Instead, deemed harm should be limited to intelligence information (as defined) and the operations, capabilities, technologies, methods and sources used to obtain or communicate that information. 

For ASIS, ASD, AGO and DIO, ‘intelligence information’ should be defined by reference to the existing definition of ‘intelligence information’ in the Intelligence Services Act. For ASIO, it should cover information obtained for the purpose of security (as defined in the ASIO Act), as well as ASIO’s foreign intelligence function in s 17(1)(e) of that Act. For ONI, the definition of ‘intelligence information’ should be linked to ONI’s statutory intelligence functions in s 7(1)(c)(d)(e) and (g) of the ONI Act. The inclusion of DIO in this offence is contingent on Recommendation 5. 

Law enforcement – electronic surveillance capabilities 

Any intentional or reckless interference with the integrity of the criminal justice system or disclosures that would undermine law enforcement capabilities are serious and deserve some type of sanction. However, it does not follow that it is necessary and proportionate for all information relating to the operations, capabilities, technologies, methods or sources of any agency with a role in enforcing the criminal law to be covered by a broad 7−10-year deemed harm offence. This covers many agencies and a very large amount of information, not all of which is particularly sensitive. 

This review recommends a cascading approach. The serious deemed harm offence in s 122.1 should apply to the technologies, capabilities and methods that support the extraordinary electronic surveillance powers that parliament has granted to a small number of agencies to combat serious crimes. These surveillance capabilities include remote computer access, telecommunications interception, network access and account takeover warrants. The specialised capabilities needed to use these powers require significant investment. They would take a long time to replace if compromised, and their loss would seriously undermine the ability to combat serious and organised crime. Other types of interference with the criminal justice system are dealt with in Recommendations 6 and 11. 

RECOMMENDATION 3: The deemed harm offences in s 122.1 should not apply to all information relating to the operations, capabilities or technologies of, or methods or sources used by, a domestic or foreign law enforcement agency. Instead, the deemed harm offence should be limited to information that relates to the technologies, capabilities and methods used to exercise special electronic surveillance powers. 

Overlapping offences for intelligence agency information 

There is almost complete overlap between agency-specific secrecy offences for ASIS, ASD, AGO, DIO, ONI and ASIO and the general offence in s 122.1. Those agency-specific offences presently apply to any information made by or for the agency in connection with its functions. For the same reasons discussed in relation to Recommendation 2, this is excessively broad and these offences should be narrowed in the way proposed by Recommendation 2. 

There is a good argument for repealing the agency-specific secrecy offences and instead relying on the Criminal Code. This would be consistent with the goal of reducing the overall number of secrecy offences. Some care in drafting would be required to ensure that all individuals presently covered by the offences (such as ASIS agents) would continue to be covered by the new offence and that new defences were not added without consideration of whether they were appropriate. The penalties are already similar. 

Most of the intelligence agencies oppose moving their general secrecy offences into the Criminal Code, primarily because they consider additional defences should not be available. For example, Part 5.6 of the Criminal Code includes defences for lawyers who provide legal advice on offences against that Part and in relation to information that has been previously published. It may complicate the drafting in the Criminal Code for some defences not to be available to people who would otherwise fall into the definition of ‘Commonwealth official’ or persons otherwise performing work for the Commonwealth. 

There are some very specific matters that apply to ASIS and ASIO that should be retained as specific offences in their own Acts. In particular, ASIS and ASIO staff and agents undertake covert intelligence collection activities that put them in positions of particular risk if their identities are disclosed. 

ASIO’s new function of assessing and granting the highest level of security vetting clearances also presents a particular risk. This information should be covered in the ASIO Act for ASIO staff and affiliates. For other Commonwealth officials, disclosure of vetting information should be covered by the harm-based offences in ss 122.2 and 122.4. 

Defence Intelligence Organisation needs statutory functions 

Currently the Criminal Code and the IS Act both describe serious criminal offences by reference to DIO’s functions. Those functions are not set in legislation; they are set by a policy document, generated by the Department of Defence, that can be changed at any time. It is not appropriate for the scope of a serious criminal offence to be changeable in this way. 

Setting out the functions of DIO in legislation or a disallowable legislative instrument would also provide for parliamentary oversight and clarity in the operation of the exceptions for DIO in the FOI Act, Privacy Act and the way DIO information is dealt with in the PID Act. If this recommendation is not accepted, I recommend that the IS Act be amended to repeal the secrecy offences for DIO and that s 122.1 of the Criminal Code not refer to DIO, as both offences are dependent on the scope of DIO’s functions. The disclosure of DIO information which is likely to harm security, defence or international relations would continue to be covered by the offence in s 122.2. 

Harm-based offence for Commonwealth officials 

The offence in s 122.2 is a 7-year (or 10-year in aggravating circumstances) harm-based offence. It applies when an official is reckless as to whether their conduct will cause harm, or is likely to cause harm, in a range of circumstances. The offence should be retained but requires increased clarity as to what is meant by ‘security’, ‘defence’ and ‘international relations’. There should be a cascading approach to law enforcement capabilities. 

The offence in s 122.2 is a 7-year (or 10-year in aggravating circumstances) harm-based offence. It applies when an official is reckless as to whether their conduct will cause harm, or is likely to cause harm, in a range of circumstances. The offence should be retained but requires increased clarity as to what is meant by ‘security’, ‘defence’ and ‘international relations’. There should be a cascading approach to law enforcement capabilities. 

‘Security’ should be defined in accordance with the definition in the ASIO Act. ‘Defence’ should be defined in a way that incorporates functions directly connected to the defence of Australia, such as those described in recommendation 2.3 of the October 2023 INSLM NSI Act Report. ‘International relations’ should cover diplomatic and military relations with foreign governments and international organisations, including bilateral and multilateral law enforcement and intelligence cooperation arrangements. 

The offence in s 122.2 should penalise disclosures that could compromise the utility of powers exercised under warrants and authorisations granted to support the investigation of crime by any agency. This includes capabilities connected to statutory powers to access information or to search people, places or things − for example, sensitive forensic capabilities used to extract information from items seized under a search warrant. Other types of interference with the criminal justice system are dealt with in Recommendations 3 and 11.

RECOMMENDATION 4: If separate general ‘deemed harm’ offences are to be retained in the Intelligence Services Act 2001, the Australian Security Intelligence Organisation Act 1979 and the Office of National Intelligence Act 2018, those offences should be narrowed so that the scope of the deemed harm is no wider than that described in Recommendation 2, except that:  for ASIS and ASIO existing specific offences relating totheidentityofcurrent and former staff, affiliates and agents should be retained.  in the ASIO Act, the offence should include a category of information connected to the function of assessing and issuing Australia’s highest level of security clearance under Part IVA of that Act. 

RECOMMENDATION 5: The functions of the Defence Intelligence Organisation should be set out in legislation or in a disallowable legislative instrument. 

The offence in s 122.2 should penalise disclosures that could compromise the utility of powers exercised under warrants and authorisations granted to support the investigation of crime by any agency. This includes capabilities connected to statutory powers to access information or to search people, places or things − for example, sensitive forensic capabilities used to extract information from items seized under a search warrant. Other types of interference with the criminal justice system are dealt with in Recommendations 3 and 11. 

RECOMMENDATION 6: The offence in s122.2 should apply to disclosures of information by officials where there is harm or likely harm to:  security, defence or international relations (as defined)  the utility of operational and technical capabilities and methods connected to statutory powers granted to any agency to access information or to search people, places or things (other than those covered by s 122.1) to combat crime  AFP protective and custodial functions and proceeds of crime functions, or  the health or safety of the Australian public or a section of the Australian public. 

‘Dealing with’ offences 

The ‘dealing with’ offences penalise conduct even when a person does not intend to disclose information. The definition of ‘deal’ is long and contains overlapping categories. A person deals with information if they receive or obtain it; collect it; possess it; make a record of it; copy it; alter it; conceal it; communicate it; publish it; or make it available. For the officials that the Commonwealth trusts to access the type of information covered by the revised ss 122.1 or 122.2, a ‘dealing with’ offence can be justified in most circumstances. With the exception of removing the initial receipt of information, this recommendation is intended to refine the definition and remove overlap, not to substantially alter the existing offence for officials. 

RECOMMENDATION 7: The definition of ‘deal with’ for the purpose of Part 5.6 should be amended so that it excludes initial receipt and does not overlap with the disclosure offences. The remaining parts of the definition (collect, possess, record and copy) are broadly justified for officials, although some clarification in drafting is suggested. 

Non-officials are in a different position from officials when it comes to assessing when criminal sanctions are justified. Non-officials do not have the same duty to protect Commonwealth information that officials do. A non-official who attempts a disclosure or commits other ancillary offences (such as inciting an official to make an unlawful disclosure) is already subject to other criminal offences. If they act on behalf of a foreign principal or for espionage, a range of criminal offences apply. Depending on the specific facts, it is possible that possession of stolen property offences may also apply. 

Considering the position of non-officials compared to officials and the range of other existing offences, it is not proportionate to apply ‘dealing with’ offences to non-officials. Disclosure offences should continue to apply as per Recommendation 12. 

Proper place of custody 

There are offences that were intended to apply to officials who remove information from a ‘proper place of custody’. These offences are not operational because, in the almost 6 years since they were enacted, regulations to define ‘proper place of custody’ have never been made. The offences are not necessary and should be repealed. 

Aggravating offences 

Currently there are 4 circumstances in which an offence by an official under ss 122.1 or 122.2 becomes an aggravated offence. Most should be repealed, as they are uncertain or arbitrary or are part of the aggravating circumstances already considered as part of sentencing, including those associated with dishonesty, gravity and scale. 

There are 2 circumstances where a penalty beyond the maximum of 7 years (for communicating) or 3 years (for dealing) is justified. The first is where an official holds the highest level of security clearance. Those with the highest level of clearance have qualitatively and quantitively different access to sensitive information than those with lower clearances or no clearance. There is also a particular harm to the Commonwealth and its relationship with foreign partners if there is a breach of trust by those granted the very highest level of security clearance. A higher maximum penalty can also be justified where a person acts with an intention or knowledge that their conduct will or is likely to cause harm. 

RECOMMENDATION 8: The offence for ‘dealing with’ information by non-officials in s 122.4A(2) should be repealed. 

RECOMMENDATION 9: The ‘proper place of custody’ offences in ss 122.1(3) and 122.2(3) should be repealed. 

RECOMMENDATION 10: The maximum penalty for offences by officials under Part 5.6 should be increased only where, at the time the person received the information or committed the underlying offence, the person held the highest level of Australian Government security clearance; or where the person intended or knew their conduct would or was likely to cause a type of harm covered by the underlying offence. 

General offence for Commonwealth officials 

The general offence for officials in s 122.4 is due to sunset in December 2024. The Attorney- General’s Department has recommended it be replaced by a new general offence. Policy and drafting work is ongoing. This recommendation can provide only in-principle guidance. 

RECOMMENDATION 11: Any general offence to replace s 122.4 should be consistent with the following principles:  The new offence should apply to disclosures that prejudice the prevention, detection, investigation, prosecution or punishment of a criminal offence against a law of the Commonwealth.  The offence should be harm-based and relate to essential public interests. However, if ‘deemed harm’ offences are to be incorporated, they should be limited to a very narrow category of information where significant harm to an essential public interest is always, or almost always, going to be the result.  The offence should cover only disclosures that cannot be adequately dealt with by existing remedies including contractual and administrative remedies.  Broad and uncertain language such as ‘functioning of government’ should be avoided.  The offence should apply to current and former Commonwealth officials and others who perform work for a Commonwealth entity in relation to information acquired in the course of their duties. However, if the scope of the offence is to be broadened it should still be closely linked to some kind of contract, agreement or arrangement with the Commonwealth.  The penalty for reckless conduct should be no more than 2 years imprisonment. If a primary justification for the new general offence is to replace specific existing offences then the offences to be repealed should be included in the legislative proposal in order to allow parliament to properly assess the necessity and proportionality of the new offence. 

Offence for non-officials 

The offence in s 122.4A applies to non-officials, including journalists. For the reasons applicable to Recommendation 1, using ‘classified information’ as an element of the offence creates significant uncertainty and real concerns with the rule of law. These are even more profound for non-officials. The remaining categories in the existing offence require revisions to bring them into line with other recommendations in this report and to ensure that the threshold for liability for non-officials is higher than for officials. 

RECOMMENDATION 12: The offence in s 122.4A for communications by non-officials should be modified so that:  classification markings do not form an element of the offence  the current requirement that actual harm be established should be maintained and the offence apply to:  causing serious damage to the security or defence of Australia, with those terms defined as per Recommendation 6  seriously undermining the utility of the technologies, capabilities and methods used to exercise special statutory powers (per Recommendations 3and4)  seriously impeding the prevention, detection, investigation, prosecution or punishment of a criminal offence against a law of the Commonwealth  prejudicing the health or safety of the Australian public or a section of the Australian public.  the maximum penalty should be approximately half the maximum penalty for a comparable communication by an official. Action should be taken to ensure that ABC and SBS staff and contractors are not inadvertently covered by the offences for officials as persons ‘otherwise engaged to perform work for a Commonwealth entity’. 

Journalism and public interest defences 

Several non-government organisations made strong arguments as to why a public interest test should be added as an element of secrecy offences and/or that a general public interest defence should be added. There is merit in these arguments. However, it would be complex and uncertain in practice. There are likely to be many competing public interests at play, some of which may shift over time. A general public interest element or defence for officials would also undermine the parliamentary intention that there be specific statutory mechanisms for providing immunity to those who bring forward wrongdoing. 

Concerns about ‘dealing with’ information by journalists, lawyers and civil society groups will be dealt with by Recommendation 8. However, if that recommendation is not accepted, additional changes to some defences will be needed. 

The current defence for disclosures in the course of reporting news and current affairs should be retained. For a journalist to invoke the defence, they must adduce or point to evidence that, taken at its most favourable, suggests a reasonable possibility that they ‘reasonably believed that engaging in the conduct was in the public interest’. This is known as an ‘evidential burden’ and is not a high bar. Once it is discharged, the prosecution must prove beyond reasonable doubt that each element of the defence does not exist. In the full context of how criminal trials operate in Australia, this approach remains reasonable. 

Nevertheless, there may be value in reframing the current ‘journalist defence’ as an exception rather than as a defence. Although the value would primarily be symbolic, it may make clearer that a reasonable belief that conduct is in the public interest is something that can be relied on by a journalist to avoid a conviction. 

Role of the Attorney-General 

The Attorney-General’s consent should be required for all prosecutions under Part 5.6. The consent requirement may not currently apply to a prosecution that proceeds summarily because consent is triggered by ‘commitment’ which occurs only for trial by indictment. Before any prosecution based on information being ‘security classified’ the Attorney- General must certify that it was appropriate that the information had a security classification. If Recommendation 1 is accepted and security classification is no longer an element of any offence then this requirement will fall away; otherwise, it should be retained. 

RECOMMENDATION 13: A new general public interest defence or element should not be added in Part 5.6. However, consideration could be given to recasting the current defence for journalists as an exception rather than a defence. 

RECOMMENDATION 14: The requirement that the Attorney-General’s consent be obtained for prosecution under Part 5.6 should be retained. The Attorney- General’s consent should be required regardless of whether the prosecution proceeds by way of committal or summary proceedings. 

Prosecution Policy of the Commonwealth 

The Prosecution Policy of the Commonwealth contains a non-exhaustive list of factors that the Commonwealth Director of Public Prosecutions (CDPP) is to take into account when considering the public interest in a prosecution. Prosecutions for secrecy offences can require consideration of the role of a free press including in exposing corruption or other wrongdoing by government officials. While the CDPP can already take this into account under the general concept of public interest, there is merit in explicitly including it in the list of public interest factors to be considered. 

RECOMMENDATION 15: Consideration should be given to revising the Prosecution Policy of the Commonwealth to expressly include the public interest in a free and open press as one of the factors to be considered in any prosecution for a secrecy offence involving a journalist or news media organisation.

Neurodiversity

'The Human Spectrum: A Critique of “Neurodiversity”'by Douglas W. Maynard in (2024) Symbolic Interaction argues 

This paper represents a sociological approach to autism spectrum disorder that critiques the terms neurodiverse and it obverse, neurotypical, because they promote a cognitive approach that mystifies what is actual and real about human activity in everyday life. It is in dynamic interactional practices rather than putative cognitive states that human diversity is manifest or observable. The empirical part of the paper, following Bleuler, defines and examines “autistic talk” as a turning away from the ordinary social world or commonsense “reality,” and engaging self-oriented practices and orientations. However, the range of participants who produce such talk is not confined to those on the putative autism spectrum. Rather, that range encompasses the entire human span. If so, then a question is raised about what autism is as a “condition,” which term individualizes social difficulties rather than appreciating that they are always about diverse social actions-in-interaction. 

Although the history of the term “neurodiversity” is well-documented, a brief review is important to the purpose of this paper, which is to suggest a sociological approach to autism in which we can see how neurodiversity and neurotypicality are tightly connected in actual and real human experience and behavior, rather than contrastively in the brain as such. For understanding autism, neurodiverse and it obverse, neurotypical, overwhelmingly promote a cognitive approach that mystifies—removes or obscures—what is actual and real about human activity in everyday life. Accordingly, it is time to take social relations seriously and appreciate how it is in lively interactional practices and not by way of putative cognitive states, whether such states are of neural or genetic origin, that human diversity is observably, witnessably manifest. This paper draws from approaches in sociology known as ethnomethodology (EM) and conversation analysis (CA) and related areas such as symbolic interactionism to explore such a social and interactional proposition. This exploration raises a question about what Autism (ASD) is as a “condition,” a matter that is addressed throughout and in the conclusion to this paper. 

In an early essay, Sinclair (1993) proposed that autism is a “way of being” that parents and others could appreciate rather than only grieve. Later in that decade, in an article for an internet mailing list called Independent Living (APA 1994; Dekker 2020), Singer (1999), a sociologist diagnosed with Asperger's Syndrome, coined the term “neurodiversity,” suggesting that the term “did not come out of the blue, but was the culmination of … experiences of exclusion and invalidation as a person.” From these origins, the thrust of “neurodiversity” usage in differing discourses relative to self-advocacy, social justice, parental groups, clinical orientations, and disability rights advocacy, has been to present “an alternative to the medical model” (Ortega 2009:74), and “to construe autism as a positive attribute” (Ortega 2009:427). However, rather than a generic alternative, critical stances relative to the medical model vary according to whether there is outright opposition, as with the “social model” of disability (Shakespeare and Watson 2002), bivalency, in which, for example, the relevance of “cure” may be accepted (although the more common attitude may be one that only embraces pharmaceutical treatment), or an orientation to autism as an identity that requires “person-first” articulation (“person with autism”)—although the latter orientation has become contested (Baron-Cohen 2017) and the current convention (Baron-Cohen 2017; Gernsbacher 2017; Vivanti 2020), followed in this paper, is to refer to the “autistic person.” 

Neurodiversity, as Kapp (2020b:2, original emphasis) puts it, includes “both neurodivergent people (those with a condition that renders their neurocognitive functioning significantly different from a ‘normal’ range), and neurotypical people (those within that socially acceptable range).” However, a difficulty here is that we do not know what a neurological “normal range” is such that it is possible to be precise about what either “typicality” or “divergence” are or could be. Until we attend to what people do, as advocated in this paper, rather than what people have (a “condition” of one kind or another), it is impossible to be precise about a continuum, and what is divergent and what is typical. A further problem is the emphasis on “people” as fixed entities or as having defined identities or conditions, rather than on action and activity and their vibrancy. Indeed, the “neuro” prefix attached to “diversity” and “typicality” derives from orientations that reflect how neuroscience has come to predominate in society—across educational, marketing, psychiatric, media, and popular endeavors (Ortega 2013:73), with a resulting “reductionist assumption” that “contemporary neuroscience can provide fairly complete and convincing explanations of who and what we are” (Racine 2010:91). Or, as Rapp (2011:25, my emphasis) has put it, “… neurodiversity is increasingly engaged as a polysemic metaphor for the acceptance of human difference lodged in the physiological brain,” such that, as Pickersgill and Van Keulen (2011:xiv) observe, both “pathology” and “normality” are now “situated within a neurological rubric.” 

Contrariwise, the field known as “critical autism studies” (Kapp 2020a; Orsini and Davidson 2013b) has developed an extensive set of considerations relating neurodiversity to issues of identity, normality, justice, uses of discourse, and possibilities for collective social action. Central to these endeavors is a challenge to “the deficit narrative autism” and any other “reductionist” approaches to autism, including those not only from a biological perspective, but also from social constructionist, cultural or other inquiry (Orsini and Davidson 2013a:13). My addition to such critical stances is to minimize what people have or are neurologically and draw further attention to an element that resides in the social things participants do and how they do them with talk and the body (cf. Maynard and Turowetz 2022; Turowetz 2015). 

As Section 1 of the paper continues, I address (A) the “social interactional subject,” (B) the contrast between common sense and autism (including autistic talk), and (C) how to gain access to the orientations of those considered to be on the autistic spectrum. Section 2 commences by (A) considering ordinary talk and its “poetics,” including (1) sound- and category-related speech productions, (2) the “whistling” of one's experience, and (3) punning, all of which have properties related to autistic talk. Also, in Section 2, we explore (B) how autistic talk can be a resource in otherwise ordinary interactions via (1) punning, (2) the de-railing of an action, and (3) controlling an agenda. The Conclusion draws implications about the autistic continuum from the empirical part of the paper.