In an era where open models are accelerating in utility OpenAI moves to platform model and offerings to lock-in the ecosystem during its first Developer Day. The platform model showcases a strategic move reminiscent of Apple’s integrated ecosystem approach. For some startups a master class in the risks associated with supply-side dependencies It’s worth watching the keynote for the new set of offerings including ChatGPT-4 Turbo, as well as the ability to create custom “agents,” called “GPTs”
“We’re introducing copyright shield. Copyright Shield means that we will step in and defend our customers and pay the costs incurred, if you face legal claims or on copyright infringement, and this applies both to ChatGPT Enterprise and the API.” IP is the unknown known, or known unknown depending on your p.o.v.
Followed by this statement. “..let me be clear, this is a good time to remind people do not train on data from the API or ChatGPT Enterprise ever.” a caution against unauthorized use of the data which could have legal, ethical, or technical implications. A signal to startups with supply-side dependencies alluded to above. We live in a complex world of dualities.
For me the feature called reproducible outputs, which ensures that every time the model runs with the same seed and inputs, it produces the same output. I have already started using gen_id seeds for creating visual continuity to test exactness and precision.
Meanwhile, traditional industry defenses are undergoing transformation (e.g. moats) In the realm of entertainment, Clear boundaries are being defined around the use of AI in the SAG-AFTRA negotiations
AMPTP addressed the issue of AI by offering an increase in salaries to professionals that allow them to be virtually replicated. It appears there is no commitment to cease training its AI systems. Wow, read that last sentence again.
The music industry, grappling with AI’s rise strategies. Among major labels Warner Music Group (WMG) hints at hopes for legislative support and a DRM Content ID-style system, while the Universal Music Group (UMG) suggests that bolstering laws, like the proposed federal right of publicity in the U.S., could address issues arising from synthetic media. (No comment -Ed)
The pace at which technology advances often outstrips the legal and business frameworks meant to govern it. An illustration of this is the emergence of AI-generated music on social platforms, such as Sorisori’s service that offers tracks mimicking well-known artists like Ariana Grande in novel contexts.
Content provenance tools, such as Nightshade and Glaze, are in their infancy. These early attempts to trace AI’s supply chain contributions face challenges in proving their efficacy.
Korean song to an AI cover of South Korean singer IU singing Cupid by K- pop girl group Fifty Fifty. For the monthly subscription fee of 14,800 won (S$15.40) on the English website, subscribers can generate up to 200 tracks of music using AI. Spot-AI-fy, a YouTube channel that specializes in AI-generated music, has a total of 235 videos on the platform. Sites like this are everywhere.
Google’s reported $2 billion investment in Anthropic, an OpenAI competitor, signifies the intensifying competition in the AI space. This is the same Anthropic currently embroiled in a lawsuit with UMG over the unauthorized distribution of copyrighted lyrics through its AI model, Claude 2. Seeking potentially tens of millions in damages and could set a legal precedent.
In journalism, the industry fresh off being destroyed by social media is moving to a block the bot and license strategy to combat the potential for AI to further lay waste to this critical field.
The license plays are just theater for the AI companies they eat, and will eat what they want. In July, The Daily Telegraph revealed allegations that Google harvested around 1m online news articles from the Daily Mail.
The recent executive order signed by President Biden addresses various facets of artificial intelligence but notably does not delve into AI’s ramifications for creative industries. Overall, the general pulse seems to prioritize ‘existential risk and other challenges, such as misinformation, safety standards, privacy, and civil rights.
IP protection is in that stack but its a long road with many issues driving at the same time. Again a differential (time/innovation) where consequence will take place. If the Library of Congress comments from Antropic indicate the companies legal stance (cited below) and if it is any indicator there is a long hard copy fight ahead.
###
Citation “Artificial Intelligence and Copyright.” Federal Register, vol. 88, no. 167, 30 Aug. 2023, Antropic comments
Citation ‘Actors’ union says no agreement on studios’ ‘final’ offer‘, Agence France-Presse (online), 7 Nov 2023
Citation Glenn CHAPMAN, ‘OpenAI Sees A Future Of AI ‘Superpowers On Demand’ – ‘, International Business Times: United Kingdom Edition (online), 7 Nov 2023
Citation ‘AI-generated music sparks debate in S. Korea’, Straits Times, The (online), 7 Nov 2023
Citation ‘Artificial Intelligence regulation starts to take shape in US and UK’, Cape Times, The (online), 7 Nov 2023
Citation: 1. UMG Music Group N.V. Q3 2023 Call
N.B. some links go to alternate sites with the same story not behind a firewall