Generative AI on Trial for Copyright Infringement - Weekly Roundup - Issue #439
This week - President Biden issues Executive Order on AI; German businesses turn to robots; the final Beatles song was released with the help of AI; "living pharmacies"; and more!
The rise of generative artificial intelligence has revolutionised various industries, but it has also sparked significant debates around copyright infringement. Those AI systems use enormous amounts of data for training, which raises questions about the originality of their output and the potential misuse of copyrighted materials. Some groups of people believe this is the case and sue AI companies for copyright infringement.
In January of this year, a group of artists sued Stability AI (the company behind a popular open-source Stable Diffusion text-to-image AI generator), Midjourney (a popular text-to-image generator built with Stable Diffusion) and DeviantArt (which released its own AI image generator based on Stable Diffusion) for copyright infringement. The artists argued that these companies used their work in training the AIs without their consent. This week, U.S. District Court Judge William H. Orrick, of the Northern District of California filed a decision in a copyright infringement class action lawsuit. All three companies filed a motion to dismiss the lawsuit and Judge Orrick largely granted it, writing “the Complaint is defective in numerous respects.”
One of the reasons the judge dismissed the case was that most of the artists’ works were not registered with the U.S. Copyright Office. However, he allowed the artists to file a new, narrower lawsuit concerning the copyrighted works.
The judge also ruled that because AI image generators reference art by many different artists when generating new images, unless it is possible to prove that the resulting image referenced solely or primarily copyrighted art and is substantially similar to that original copyrighted work, it is likely not infringing on the original work.
As a poster child for the generative AI revolution, OpenAI is involved in many lawsuits (at least 10 lawsuits according to this list).
In November 2022, Microsoft, Github, and OpenAI got sued for Copilot - Github’s AI programming assistant. The lawsuit alleges that Copilot - Github’s AI assistant for software engineers - violates copyright law by reproducing open-source code. The lawsuit alleges that Copilot - Github’s AI assistant for software engineers - violates copyright law by reproducing open-source code. The website promoting the lawsuit states that Copilot “relies on unprecedented open-source software piracy.” The lawsuit targets Github (which offers Copilot), OpenAI (whose Codex model powers Copilot) and Microsoft (the owner of Github and also OpenAI’s largest partner). In May 2023, the judge upheld claims that Codex's ability to reproduce code breaches software licensing terms and that Copilot and Codex reproduce copyrighted code without essential copyright management information. However, he dismissed several other allegations.
OpenAI is also being sued by authors. John Grisham, Jodi Picoult, George RR Martin, and 17 other authors are suing OpenAI for “systematic theft on a mass scale.” The lawsuit, filed by the Authors Guild in New York, alleges “flagrant and harmful infringements of plaintiffs’ registered copyrights” and labels ChatGPT a “massive commercial enterprise” that relies upon “systematic theft on a mass scale.” This is the second lawsuit against OpenAI filed by writers. Previously, a group of authors, including Sarah Silverman, filed a similar lawsuit in California.
The main challenge for groups suing AI companies is proving that their copyrighted work was used in training the AI, that the AI predominantly uses copyrighted work to answer a prompt, and that the response is very similar to the said copyrighted work. Judge Orrick’s ruling suggests that this might be the case. The AI companies don't want to disclose their training datasets as doing so might reveal which copyrighted works went into training their AI systems, potentially exposing them to copyright lawsuits. However, when the EU AI Act comes into force, they might have to do just that if they want to operate within the EU. We will see how the legal battles around copyrighted material and AI evolve when that happens.
If you enjoy this post, please click the ❤️ button or share it.
Becoming a paid subscriber is the best way to support the newsletter.
If you enjoy and find value in my writing, please hit the like button and share your thoughts in the comments. Share the newsletter with someone who will enjoy it, too.
You can also buy me a coffee if you enjoy my work.
🦾 More than a human
Some deaf children in China can hear after gene treatment
Chinese researchers have developed a gene therapy that restored hearing in children. This therapy aims to correct a gene defect that prevents the body from producing a protein called otoferlin, essential for transmitting sounds to the brain. The study successfully restored hearing in four children. The therapy was quick, too. As the mother of one of her children says, her child was hearing with the treated ear for the first time in less than a month. Researchers hope their results will open a new range of affordable gene therapies targeting eye and ear diseases.
▶️ How sports medicine can maximize the longevity runway (12:09)
Here is a conversation with Anna Erat, the Medical Director at Longevity Center Zurich, about how knowledge of optimizing top athletes' performance can be applied to longevity.
“Living pharmacies” could mean you never forget to take your meds again
Imagine a device that, once implanted into the human body, produces medicine indefinitely without the need for refills or replacements. That's the project DARPA is currently developing. Named NTRAIN, this microchip implant is designed to contain cells engineered to release peptides that naturally control our circadian clock, activated by a light on the implant. DARPA envisions that this device could help soldiers quickly overcome jet lag or adjust to changes in their work schedules. In-human studies of the NTRAIN device are expected to begin in 2025. If successful, these devices could revolutionise treatment for many, such as those with diabetes, by continuously releasing insulin or other medications.
🧠 Artificial Intelligence
President Biden Issues Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence
President Biden has issued a long-anticipated American approach to regulating the AI industry. The executive order sets out to establish new standards for AI safety and security, safeguard Americans' privacy, advance equity and civil rights, foster innovation and competition, and promote American leadership, among other objectives. It addresses various facets of the AI industry, from defining computing thresholds to streamlining visas for AI talent. At the same time, President Biden unveiled AI.gov, a new US government website showcasing the federal government’s approach to AI and AI safety. If you want to learn more about what exactly the new executive order contains and don't have the time to go through all 111 pages of it, I recommend checking out this summary written by
At the Direction of President Biden, Department of Commerce to Establish U.S. Artificial Intelligence Safety Institute to Lead Efforts on AI Safety
The Biden-Harris Administration announced plans for the U.S. Department of Commerce, through the National Institute of Standards and Technology (NIST), to establish the U.S. Artificial Intelligence Safety Institute (USAISI). This institute will lead the government's AI safety and trust initiatives, aiming to develop standards for AI model safety, security, and testing, including measures for authenticating AI-generated content.
What the U.N.’s AI Advisory Group Will Do
The UN has introduced a new advisory body focused on building consensus regarding the challenges presented by AI and the potential of international collaboration to address them. Although the body itself holds little power, its guidance could shape the structure of a future UN agency overseeing AI governance. The UN Secretary-General António Guterres emphasized the critical need for multidisciplinary, global discussions on AI governance to maximize its benefits and minimize risks. The advisory group, consisting of 39 members from diverse backgrounds and sectors, is part of a broader international effort towards AI governance, including initiatives like the UK AI Safety Summit and the G7 AI code of conduct.
▶️ Shane Legg (DeepMind Founder) - 2028 AGI, New Architectures, Aligning Superhuman Models (44:18)
In this interview, Shane Legg, co-founder of DeepMind, shares his thoughts and ideas on AGI. The conversation touches on topics such as how to measure AGI and what is missing from today’s metrics; strategies for ensuring superhuman models are ethical and safe; the need for new architectures in AGI development; his prediction for AGI's emergence around 2028 and the rationale behind it; and why he believes multimodality will be the next significant milestone.
The employees secretly using AI at work
A recent survey of 2,000 global IT decision-makers revealed that 75% are either considering or have already implemented bans on ChatGPT and other generative AI applications in the workplace. Of these, 61% intend for these measures to be long-term or permanent. The primary concern for employers is the risk of data leaks and the unintentional training of AI with confidential information. Despite potential bans, employees are finding covert ways to access and use these tools. Many of them see these tools as incredibly beneficial, as they ease their workload and improve efficiency. As AI becomes an integral part of the workplace, restricting its use could lead to employee dissatisfaction and may prompt them to seek employment where they have access to the tools they find beneficial.
The Beatles released their final song with the help of AI
More than 50 years after their breakup, The Beatles have released their final song. The song - Now And Then - came from unreleased material recorded before John Lennon’s death. It remained unpublished earlier because of a persistent buzzing sound in the recording. However, with the aid of AI, it was possible to isolate Lennon's clear voice and complete the track. Alongside the new song, The Beatles released a short documentary on how their final song came to be.
🤖 Robotics
As baby boomers retire, German businesses turn to robots
In June, Germany had approximately 1.7 million unfilled jobs, a situation attributed to the retirement of baby boomers and a smaller incoming cohort of new workers. This trend is expected to accelerate, as the Federal Employment Agency expects the pool of workers to shrink by 7 million people by 2035. To address the shortfall, German companies are increasingly relying on robots. Heavy investment in automation by car makers and other industrial giants means Germany is already the world's fourth-biggest market for robots, and the largest in Europe. Moreover, it's not just major industries that are embracing automation. Small and medium-sized businesses, including bakeries, laundries, and supermarkets, are also considering robots to address their staffing needs.
🧬 Biotechnology
CRISPR cure for HIV now tested in 3 people
Excision BioTherapeutics is among several groups investigating the potential of CRISPR gene-editing technology to cure HIV directly. After encouraging results in studies involving mice and non-human primates with SIV — a simian counterpart to HIV — Excision launched a clinical trial for its CRISPR-based HIV treatment in 2022. On October 25, 2023, they presented the initial data from this trial at the European Society for Gene & Cell Therapy’s annual meeting in Brussels, Belgium. Notably absent from the announcement was any information on whether the therapy worked as the primary objective of the trial was to test the safety of the treatment. Excision confirmed that none of the patients encountered severe adverse effects due to the CRISPR therapy. They plan to release more comprehensive trial data in 2024.
Gene Transfer Leads to Longer Life and Healthspan
The naked mole rat is known for its remarkable resistance to ageing and cancer. Researchers at the University of Rochester wondered what would happen if genes responsible for the naked mole rat’s abilities were transferred to another species. They did just that by transferring genes that produce high-molecular-mass hyaluronic acid (HMM-HA), a compound believed to be linked to the naked mole rat’s longevity, into mice. These modified mice exhibited a reduced cancer rate and a longer lifespan compared to control mice. While this breakthrough demonstrates that genes from one species can positively influence another species' longevity, its direct application in humans remains complex and uncertain.
Thanks for reading. If you enjoy this post, please click the ❤️ button or share it.
Humanity Redefined sheds light on the bleeding edge of technology and how advancements in AI, robotics, and biotech can usher in abundance, expand humanity's horizons, and redefine what it means to be human.
A big thank you to my paid subscribers, to my Patrons: whmr, Florian, dux, Eric, Preppikoma and Andrew, and to everyone who supports my work on Ko-Fi. Thank you for the support!
You can follow Humanity Redefined on Twitter and on LinkedIn.
Some excited to have found your Stack! Big fan of the subjects you’re covering here. I am writing a Science IRL/FIction mashup this month about gene editing and anti-aging so it’s a special treat reading this post about the latest CRISPR and gene therapy successes. Feel free to keep me honest, eh? :) Looking forward to your next roundup!
"The judge also ruled that because AI image generators reference art by many different artists when generating new images, unless it is possible to prove that the resulting image referenced solely or primarily copyrighted art and is substantially similar to that original copyrighted work, it is likely not infringing on the original work."
ALL art produced by humans, whether art, literature, music...etc is "learned" from someone else's work, yet one would never think of suing an artist for painting a bowl of fruit just because someone else had done so in the past. On the surface, seems like a frivolous suit.