Federal judge rules AI art can't be copyrighted

Welcome to The Dispatch! We are the newsletter that keeps you informed about AI. Each weekday, we scour the web to aggregate the many stories related to artificial intelligence; we pass along the news, useful resources, tools or services, guides, technical analysis and exciting developments in open source.

In today’s Dispatch:

• A potential lawsuit by The New York Times against OpenAI for copyright infringement in training ChatGPT could end up costing OpenAI enormously. NPR theorizes that the company could be ordered to rebuild ChatGPT's dataset from scratch and pay fines of up to $150,000 for each piece of infringing content. The case could also trigger similar lawsuits from other publishers and force OpenAI to defend its fair use claims in court.

• A new study finds AI bots can solve CAPTCHAs for identifying objects or distorted text more accurately and quickly than people. The CAPTCHA approach to blocking automated attacks is likely becoming obsolete. Designing a new system could prove challenging, as there’s a ‘built-in ceiling’: make a system too difficult to solve, and users will give up.

• Meta is reportedly close to releasing ‘Code LLaMa’ - an open source AI focused on generating code. Code LLaMa is built on Meta's LLaMa-2 language model and will suggest code to developers as they type, competing with proprietary offerings from OpenAI, Google, and Microsoft. This aligns with Meta's continued strategy of releasing free AI tools to support custom AI development. Code LLaMa could be released as soon as this week.

Plus: A federal judge rules against AI-generated art, big advertisers betting on AI, a Chinese robot with enough precision to peel raw eggs, trending tools and more!

Judge rules AI art can’t be copyrighted

From Bloomberg Law: A US federal judge has ruled that AI-generated artwork cannot be copyrighted, stating that human authorship is required for copyright protection. The case involved Stephen Thaler's attempt to copyright an image created by an AI he developed, after he had been rejected multiple times by the US Copyright Office.

More details:

• Judge Beryl A. Howell presided over the lawsuit against the US Copyright Office. Thaler tried multiple times to copyright the image as a "work-for-hire" with himself as the owner, but was repeatedly rejected. The judge stated the case was ‘not so complex’ because Thaler admitted in his application that he played no role in creating the work.

• In her ruling, Judge Howell stated that copyright has never been granted for work without a "guiding human hand" and that "human authorship is a bedrock requirement of copyright." She acknowledged that AI will raise new questions about how much human input is required for copyright as models are often trained on existing work.

• The ruling is the first in the country to establish a boundary on the legal protections for AI-generated artwork. Thaler plans to appeal.

Takeaways: At some point there will have to be a legal clarification on what constitutes ‘human authorship’ when the output is not solely AI-generated. For example, what if the author had made a very simple edit to the AI-generated photo? Is that all that would be required for the human element? Probably not - but the law doesn’t say. In February, the US Copyright Office argued that when generative AI is involved, at least a ‘modicum’ of human input is required. That’s a low bar; but under what circumstances full copyright protections might be extended to AI/human collaborative projects remains to be seen.

Why the great ‘AI backlash’ came for a tiny startup you’ve likely never heard of

From WIRED (may be paywalled): More data-scraping controversy: a small literary analytics startup called Prosecraft has been shut down after backlash from authors went viral. Prosecraft was doing computational analysis of writing styles and language use in books. It provided stats on things like how vivid or abstract the writing was; an author could see how their book stacked up against others in areas like vocabulary complexity. But authors were angry it used pirated copies without permission to build its database.

More details:

• Prosecraft’s use of scraped data reflects a common practice by AI companies that is now facing widespread backlash. The reaction shows growing wariness about unauthorized uses of creative work.

• Prosecraft's founder never saw the controversy coming. Benji Smith was a software engineer who built the site as a side project, hoping it would be useful for authors. Prosecraft was free to use.

• When authors found their books analyzed on the site without consent, many reacted angrily on social media. Prosecraft received hundreds of cease and desist letters within 24 hours.

• Experts say Prosecraft may have had a fair use defense if sued, but its optics are bad in the current climate of artists demanding more control over AI training.

Takeaways: One publishing industry analyst sees the backlash against Prosecraft as misguided - even using the term “shrieking hysteria” to describe it. It’s clear that we’re entering a new era where creators are fiercely protective of their intellectual property. Prosecraft was created with genuinely good intentions; it didn’t matter - the backlash was too severe.

From our sponsors:

Calendar.com

Calendar is a productivity tool from the company of the same name that helps users solve calendar problems by helping to save time, be more productive, and focus on the things that matter most!

Trusted and used by over 80,000 companies.

Try free for 14 days

Social media/Video/Podcast:

• LK-99: meet a materials scientist who put the room-temperature superconductivity claim to the test [Podcast]

• 5 Mind blowing AI tools every researcher should know about but doesn't [YouTube]

• Adobes AI 'FIREFLY Photoshop' has everyone stunned! (how to use Firefly) [YouTube]

• (Discussion) Artificial Intelligence Will Entrench Global Inequality - The debate about regulating AI urgently needs input from the global south. [Reddit]

• Chinese robot using high precision to peel a raw egg [X]

Did you know? 

Artificial intelligence is being used to discover new anti-aging medicines. Researchers at the University of Edinburgh recently fed a machine learning algorithm examples of known drugs that can slow aging. The AI model was able to analyze thousands of molecules in minutes and identify 21 top candidates that could potentially be new senolytics - drugs that eliminate damaged, inflammatory cells to combat age-related diseases. Of those 21, the researchers experimentally confirmed 3 novel senolytic compounds.

Trending AI Tools & Services:

• Consensus: Use AI to find insights in research papers

• Flythrough by Luma Labs: Create 3D virtual tour of your space with NeRFs as easily as recording a video

• Human Generator: Generated humans are produced by AI based on the millions of legally-sourced images seen by the model.

• Deasie: Data governance for language model applications.

• ChartPixel: Turn your raw data into charts and written insights in 30 seconds