Can Artificial Intelligence software claim copyright in any work that was created using their algorithms?
The short answer is “no”, since only humans can establish copyright in original creative works. Copyright can be assigned to a company or trust, or it can be created under various forms of creative commons, but there still needs to be a human author behind the copyright material. While copyright may lapse over time, it then becomes part of the public domain.
However, the extent to which a human author can claim copyright in a work that has been created with the help of AI is now being challenged. A recent case in the USA has determined that the author of a graphic novel, which included images created using Midjouney, cannot claim copyright in those images. While it was accepted that the author devised the text and other prompts that the software used as the generative inputs, the output images themselves could not be the subject of copyright protection – meaning they are either in the public domain, or they fall under some category of creative commons? This case also indicates that, in the USA at least, failing to declare the use of AI tools in a work when applying for copyright registration may result in a rejected application.
Does this decision mean that the people who write AI programmes could claim copyright in works created using their software? Probably not – as this would imply that Microsoft could establish copyright in every novel written using Word, especially its grammar and spelling tools.
On the other hand, programmers and software developers who use copyright material to train their models may need to obtain relevant permission from the copyright holders (as would anyone using the AI tools and who uses copyright content as prompts), unless they could claim exemptions under “fair dealing” or “fair use” provisions.
Over the past few years, there have been a significant number of serious data breaches among among banks, utilities, telcos, insurers and public bodies. As a result, hackers are able to access the confidential data and financial records of millions of customers, leading to ransomware demands, wide dissemination of private information, identity theft, and multiple phishing attempts and similar scams.
What most of these hacks reveal is the vulnerability of centralised systems as well as the unnecessary storage of personal data – making these single points of failure a target for such exploits. Worse, the banks and others seem to think they “own” this personal data once they have obtained it, as evidenced by the way they (mis)manage it.
I fully understand the need for KYC/AML, and the requirement to verify customers under the 100 Points of Identification system. However, once I have been verified, why does each bank, telco and utility company need to keep copies or records of my personal data on their systems? Under a common 100 Points verification process, shouldn’t we have a more efficient and less vulnerable system? If I have been verified by one bank in Australia, why can’t I be automatically verified by every other bank in Australia (e.g., if I wanted to open an account with them), or indeed any other company using the same 100 Points system?
Which is where the concept of Self-Sovereign Identity comes into play. This approach should mean that with the 100 Points system, even if initially I need to submit evidence of my driver’s license, passport or birth certificate, once I have been verified by the network I can “retrieve” my personal data (revoke the access permission), or specify with each party on the network how long they can hold my personal data, and for what specific purpose.
This way, each party on the network does not need to retain a copy of the original documents. Instead, my profile is captured as a digital ID that confirms who I am, and confirms that I have been verified by the network; it does not require me to keep disclosing my personal data to each party on the network. (There are providers of Digital ID solutions, but because they are centralised, and unilateral, we end up with multiple and inconsistent Digital ID systems, which are just as vulnerable to the risk of a single point of failure…)
But of course, banks etc. insist that not only do they have to ask for 100 Points of ID each and every time I open an account, they are required to retain copies or digital versions of my personal data. Hence, we should not be surprised by the number of data hacks we keep experiencing.
The current approach to identity in banking, telcos and utilities is baffling. Just a few examples I can think of:
1. In trying to upgrade my current mobile phone plan with my existing provider, I had to re-submit personal information via a mobile app (and this is a telco that experienced a major hack last year, resulting in me having to apply for a new driver’s license). If I have already been verified, why the need to ask for my personal data again, and via a mobile app?
2. I’ve lived at my current address for more than 5 years. I still receive bank statements intended for the previous occupant. I have tried on numerous occasions to inform the bank that this person is no longer living here. I’ve used the standard “Return to Sender” method, and tried to contact the bank direct, but because I am not the named account addressee or authorised representative, they won’t talk to me. Fair enough. But, the addressee is actually a self-managed superannuation fund. Given the fallout from the Banking Royal Commission, and the additional layers of verification, supervision and audit that apply to such funds, I’m surprised that this issue has not been picked up the bank concerned. It’s very easy to look up the current registered address of an SMSF via the APRA website, if only the bank could be bothered to investigate why the statements keep getting returned.
3. I have been trying to remove the name of a former director as a signatory to a company bank account. The bank kept asking for various forms and “proof” that this signatory was no longer a director and no longer authorised to access the account. Even though I have done this (and had to pay for an accountant to sign a letter confirming the director has resigned their position), if the bank had bothered to look up the ASIC company register, they would see that this person was no longer a company officer. Meanwhile, the bank statements keep arriving addressed to the ex-director. Apparently, the bank’s own “systems” don’t talk to one another (a common refrain when trying to navigate legacy corporate behemoths).
In each of the above, the use of a Digital ID system would streamline the process for updating customer records, and reduce the risk of data vulnerabilities. But that requires effort on the part of the entities concerned – clearly, the current fines for data breaches and for misconduct in financial services are not enough.
Last week’s Jobs and Skills Summit hosted by the Federal Government in Canberra was clearly designed to be a statement of intent by Prime Minister Anthony Albanese and his Labor administration. Part policy endorsement, part policy road map, the Summit was hailed (by the Prime Minister at least) for reaching agreement on “36 immediate initiatives”. By all accounts, it was a jolly affair and everyone in the Government sounded very pleased with themselves. The reality is that despite some significant pronouncements, most of them lack detail, many of them relate to existing initiatives, a number of the “36 agreements” were largely concluded and/or telegraphed ahead of the Summit – and of course, the one item that got most attention was the most divisive: the renewed prospect of multi-employer collective bargaining.
Number of Australian companies by employment size, 2018-2022 (Source: ABS)
There were some contentious views about the small business association’s pre-Summit MoU with the ACTU. Some peak industry bodies and other commentators felt that COSBOA had “sold out” in apparently agreeing to sector-wide negotiations on pay and conditions. However, this does not appear to be the case – COSBOA is merely seeking better co-operation and consultation on areas of mutual interest, and is not endorsing any form of enforced unionisation or compulsory sector bargaining. There have been suggestions that sector-wide collective bargaining will result in higher wages, but without more detail, and pending greater clarity on the “Better Off Overall Test”, this will simply add friction to the current debate about wage and employment growth.
If we do return to a previous form of Industrial Relations policy, it’s interesting to look at the latest ABS data on Australian businesses by employment size (table above). I think it’s worth noting the number of working people in Australia who are employed by SMEs. Large employers are actually small in number, so if multi-employer collective bargaining does come into effect, it could mean tens of thousands of businesses will be involved, and many probably for the first time. On the other hand, in an industry like construction, which is both highly unionised and covered by significant industry awards, many workers are either self-employed or they are employed by independent sub-contractors.
Representation at the summit was reasonably well-balanced, between Unions (including Industry Superfunds), Business (individual companies and industry associations), the NFP and Community sectors, Academia, Think Tanks, and of course Politics. The absence of the Leader of the Federal Opposition meant that his voter base was effectively disenfranchised, although his Deputy (and Leader of the National Party) did attend. Go figure.
Much was said about “streamlining” and “updating” parts of the Industrial Relations regime. Like Australia’s tax laws, the system of Modern Awards as overseen by the Fair Work Commission feels unwieldy, unnecessarily complex, over-bureaucratic, at times vague, and often archaic bordering on arcane. There are currently over 140 different awards in place – some of them relate to an individual company, some to a particular trade or profession, and some cover a whole industry. Interpretation is often in the eye of the beholder as to whether or not it applies to a particular employer and/or employee – here is an extract from one award:
“NOTE: Where there is no classification for a particular employee in this award it is possible that the employer and that employee are covered by an industry modern award or a modern award with occupational coverage.” (Emphasis added.)
I should add that one reason given by the Labor Government for removing the prohibition on sector-wide collective bargaining is because the process for employers to request an exemption from the relevant Minister is “too cumbersome”. I don’t see how this is so given that much of the IR system is overly bureaucratic. Surely the reason for this administrative process is to avoid collusion and other cartel-like activities that would otherwise fall foul of competition law and anti-trust provisions.
The Summit had some notable things to say about gender equality and pay parity, (“Legislate same job, same pay”), training, immigration and child care; but some proposals sound vague without defined objectives (“Boost quantum technology research and education”); draconian if they inhibit workplace flexibility, especially in seasonal industries (“Limit the use of fixed-term contracts”); or too aspirational without more detail such as specific goals and measurable targets (“Leverage greater private capital into national priority areas, including housing and clean energy”). We know that Labor ministers have been vocal in their dislike of the so-called “gig economy” (a “cancer” on the economy, and “I’d like to regulate the sh*t out of it”), but perhaps they need to do more to understand why some workers actually prefer it, and what benefits it brings in terms of workplace flexibility, especially in start-ups and emerging sectors, many of which are SMEs from where much of our longer-term innovation and employment opportunities actually come.
One item that didn’t receive as much attention was the “Digital Apprenticeships Scheme”, which (subject to details…) would likely have the combined support of the Tech Council of Australia and the ACTU. Certainly, despite a vibrant and innovative IT sector, and some notable high-tech and high-end manufacturing businesses in Australia, we lag behind in STEM education, and lack basic digital literacy skills in the wider population. (Hence the need for adjustments to the skilled migration scheme?) A friend of mine who runs a small manufacturing business in Melbourne recently hired an Office Assistant. The successful candidate claimed to be proficient in standard productivity tools such as Word and Excel. In fact, they didn’t know how to COPY-PASTE, nor how to use the SUM-ALL function, which are both very basic routines. They thought they could “wing it” by watching a YouTube video…
Finally, if there is one note of caution or concern about the Summit, it is the niggling thought that this was more of a talk-fest, and that any new ideas to have emerged were either covered by existing programmes and “policy settings”, or were already in train. Going through the list of Outcomes, I counted at least three dozen separate initiatives (Plans, Schemes, Agreements, Reports, Statements, Codes, Programs, Compacts, Task Forces, Working Groups or Funds) many of which already exist, or were part of Labor’s election promises, or have been proposed prior to the Summit. (And that list excludes Federal Ministries and Government Departments.) Sounds a lot like “Talks about Talks”, with “new” money already allocated and spoken for (hence Labor’s push back on some of the implied costs of the Summit proposals). At worst, this “wish list” represents a huge amount of expensive and bureaucratic overlay, whereas we need agile and flexible economic, education and employment measures.
1. The paradoxical corollary to Moore’s Law on the rate of increase in computing power is Murphy’s Moore’s Law: that in striving for improved performance and perfectionism in all things digital, equally we risk amplifying the limitations inherent in analog technology. in short, the more something improves, the more it must also get worse. (See also my previous blogs on the problem of digital decay, and the beauty of decay music.)
2. In the realm of digital music and other platforms (especially social media), stripping out the noise (to leave only the signal) results in an impoverished listening, cultural and social experience; flatter sound, less dynamics, narrower tonal variation, limited nuance, an absence of context. In the case of streaming music, we lose the physical connection with the original artwork, accompanying sleeve notes, creative credits and even the original year of publication.
Thinking about #1 above, imagine this principle applied to #AI: would the pursuit of “digital perfectionism” mean we lose a large part of what makes analogue homo sapiens more “human”? Would we end up compressing/removing “noise” such as doubt, uncertainty, curiosity, irony, idiosyncrasies, cognitive diversity, quirkiness, humour etc.?
As for #2, like the author, I’m not a total Luddite when it comes to digital music, but I totally understand his frustration (philosophical, phonic and financial) when discussing the way CDs exploit “loudness” (in the technical sense), how .mp3 files compress more data into less space (resulting in a deterioration in overall quality), and the way streaming platforms have eroded artists’ traditional commercial return on their creativity.
The book also discusses the role of social media platforms in extracting value from the content that users contribute, reducing it to homogenised data lakes, selling it to the highest bidder, and compressing all our personal observations, relationships and original ideas (the things that make us nuanced human beings) into a sterilsed drip-feed of “curated” content.
In the narrative on music production, and how “loudness” took hold in the mid-1990s, Krukowski takes specific aim at the dreaded sub-woofer. These speakers now pervade every concert, home entertainment system, desk-top computer and car stereo. They even bring a distorted physical presence into our listening experience:
“Nosebleeds at festivals, trance states at dance clubs, intimidation by car audio…. When everything is louder than everything else, sounds lose context and thus meaning – even the meaning of loud.”
The main issue I have with digital music is that we as listeners have very little control over how we hear it – apart from adjusting the volume. So again, any nuance or variation has been ironed out, right to the point of consumption – we can’t even adjust the stereo balance. I recall that my boom box in the 1980s had separate volume controls for each speaker, and a built-in graphic equalizer. To paraphrase Joy Division, “We’ve Lost Control”.
Content in Context 