AI Assisted Scams: Cautioning the Artistic World and Highlighting Protective Mechanisms
By Nico Hahlos, USyd placement student, Arts Law
The increasing sophistication of AI has amplified risk for artists, who now face a heightened likelihood of facing large-scale intellectual property appropriation; functioning to destabilise the creative economy, and endangering the livelihoods of artists. Certainly, AI has reared its ugly head for artists. What was once forwarded as an innovative extension of creative tooling, has now emerged as a disruptive force in the artistic environment. Indeed, rather than facilitating artistic production, AI has now destabilised the economic and moral groundings of authorship itself. This is observable in two distinct risks posed by its development. Firstly, AI companies rely on the unpermitted use of creative work- whether literature, music, or art- as an educational base for their own systems: ‘teaching’ it how to generate similar content. Secondly, it has facilitated AI-driven scams targeting independent artists. As will be delineated, this has produced disastrous consequences, whereby legal enforcement is increasingly difficult to attain, and the reputational harm is immense.
Self-sustaining theft
At a systemic scale, generative AI has assisted the mass collection of vast datasets comprised of art, images, text, and music; conducted without the consent, recognition or renumeration of impacted creators. Problematically, these practices are frequented by large corporate technology companies, who frame such uses as being fair. The resulting outputs however, have been observed to compete with the very artists whose property has been harvested. Indeed, this blurring of the boundary between inspiration and appropriation has functioned to undermine traditional intellectual property protections, while destabilising creative markets as demand for original production is curbed. A key example of this comes from Spotify, which has been noted to host AI generated versions of existing songs, produced by legitimate musicians. These often imitate musical elements unique to these artists: style, voice, and composition. The problem is multifaceted. Financially, in the Spotify example, streams and royalites would be diverted from the human artists, impacting their livelihoods. Beyond fiscal returns, such practices illuminate an institutional asymmetry: whereby creators bear the cost of ‘production,’ and corporations siphon the economic benefits. This theft is particularly problematic on an extrapolated scale, wherein accumulated cultural identities, and signatures of artists are being commodified to reinforce the market dominance of technology firms.
Supercharging predatory scams
Beyond systemic creative theft, generative AI has amplified the efficacy of predatory scams, whereby AI has produced highly personalised, and realistic targeting schemes for authors and artists. In effect, this has complicated identifying the traditional red flags characteristic of such schemes, as AI-communications are often designed to mimic legitimate industry actors. Such communications are easily formulated- wherein AI can efficiently comb through author biographies, social media activity, and publication histories- and hyper-personalised to the aspirations of the individual. This has been a change of tactics for scams: moving beyond volume to psychological manipulation.
A key example of this involved debut novelist Jon Cock, who was rorted of $10,000 by a network of scammers utilising AI-generated correspondence. Credible expressions of interest in his work were followed by fraudulent demands for author licenses, and marketing packages: exploiting the clear lack of experience in junior artists. As noted, this form of “relationship fraud” does not represent a one-off extraction of credit card details but involves a sophisticated operation extending weeks or months. These tactics are easily scaled, carried out, and completed: improving success rate while maintaining low operational effort.
The perceived legitimacy of these operations not only stems from the building of personal relationships, but via the creation of deceptive corporate facades. For example, investigations conducted into a network publishers- notably that of ‘Melbourne Book Publisher’- revealed the use of cloned websites and generated staff profiles. These operations often favour impersonating industry giants such as Penguin Random House; serving as a ‘honey trap’ for aspiring writers. Again, the reproduction of proven and identifiable branding reduces operational effort while maintaining the perception of legitimacy.
Shortcomings of the legal field
Legal institutions often face significant hurdles in regulating these operations, as current copyright frameworks appeal to an older form of theft. The struggle to address models drawing from millions of sources at unprecedented speed, is exacerbated by how easily such operations are reproduced. For example, where a fraudulent site is removed, it is easily replaced under a new domain. Enforcement mechanisms are further hampered by established business advertising these schemes on their websites: also fuelling the perceived legitimacy of such scams. As such, these uses of AI pose an existential threat to the creative field; more so than ever requiring vigilant practices from artists, and a shift away from ‘reactive’ punishment.
