Featured Article : New Certification For Copyright Compliant AI

Following many legal challenges to AI companies about copyrighted content being scraped and used to train their AI models (without consent or payment), a new certification for copyright-compliant AI has been launched.

The Issue 

As highlighted in the recent case of the New York Times suing OpenAI over the alleged training of its AI on New York Times articles without permission for free (with the likelihood of a ‘fair use’ claim in defence), how AI companies train their models is now a big issue.

The organisation ‘Fairly Trained’ says that its new Licensed Model certification is intended to highlight this difference between AI companies who scrape data (and claim fair usage) and AI companies who license it, thereby getting permission and pay for training data (i.e. they choose to do so for ethical and legal reasons). As Fairly Trained’s CEO, Ed Newton-Rex says: “You’ve got a bunch of people who want to use licenced models and you’ve got a bunch of people who are providing those. I didn’t see any way of being able to tell them apart” 

Fairly Trained says it hopes its certification will “reinforce the principle that rights-holder consent is needed for generative AI training.” 

Fairly Trained – The Certification Initiative

The non-profit ‘Fairly Trained’ initiative has introduced a Licensed Model (L) certification for AI providers that can be obtained by (awarded to) any generative AI model that doesn’t use any copyrighted work without a licence.

Who? 

Fairly Trained says the certification can go to “any company, organisation, or product that makes generative AI models or services available” and meets certain criteria.

The Criteria  

The main criteria for the certification includes:

– The data used for the model(s) must be explicitly provided to the model developer for the purposes of being used as training data, or available under an open license appropriate to the use-case, or in the public domain globally, or fully owned by the model developer.

– There must be a “robust process for conducting due diligence into the training data,” including checks into the rights position of the training data provider.

– There must also be a robust process for keeping records of the training data that was used for each model training.

The Price 

In addition to meeting the criteria, AI companies will also have to pay for their certification. The price is based on an organisation’s annual revenue and ranges from $150 submission fee and $500 annual certification fee for an organisation with a $100k annual revenue to a $500 submission fee and $6,000 annual certification fee for an organisation with a $10M annual revenue.

What If The Company Changes Its Training Data Practices? 

If an organisation acquires the certification and then changes its data practices afterwards (i.e. it no longer meets the criteria), Fairly Trained says it is up to that organisation to inform Fairly Trained of the change, which suggests that there’s no pro-active checking in place. Fairly Trained does, however, say it reserves the right to withdraw certification without reimbursement if “new information comes to light” that shows an organisation no longer meets the criteria.

None Would Meet The Criteria For Text 

Although Fairly Trained accepts that its certification scheme is not an end to the debate over what creator consent should look like, the scheme does appear to have one significant flaw at the moment.

As Fairly Trained’s CEO, Ed Newton-Rex has acknowledged, it’s unlikely that any of the major text generation models could currently get certified because they have been trained upon a large amount of copyrighted work, i.e. even ChatGPT is unlikely to meet the criteria.

The AI companies argue, however, that they have had little choice but to do so because copyright protection seems to cover so many different things including blog and forum posts, photos, code, government documents, and more.

Alternative? 

Mr Newton-Rex has been reported as saying he’s hopeful that there will be models (in future) that are trained on a small amount of data and end up being licensed, and that there may also be other alternatives. Examples of some ways AI models could be trained without using copyrighted material (but probably not without consent) include:

– Using open datasets that are explicitly marked for free use, modification, and distribution. These can include government datasets, datasets released by academic institutions, or datasets available through platforms like Kaggle (provided their licenses permit such use).

– Using works that have entered the public domain, meaning copyright no longer applies. This includes many classic literary works, historical documents, and artworks. Generating synthetic data using algorithms. This could include text, images, and other media. Generative models can create new, original images based on certain parameters or styles (but could arguably still allow copyrighted styles to creep in).

– Using crowdsourcing and user contribution, i.e. contributions from users under an open license.

– Using data from sources that have been released under Creative Commons or other licenses that allow for reuse, sometimes with certain conditions (like attribution or non-commercial use).

– Partnering / collaboratiing with artists, musicians, and other creators to generate original content specifically for training the AI. This can also involve contractual agreements where the rights for AI training are clearly defined.

– Using web scraping but with strict filters to only collect data from pages that explicitly indicate the content is freely available or licensed for reuse.

Collaboration and Agreements 

Alternatively, AI companies could choose to partner with artists, musicians, and other creators to generate original content (using contractual agreements) specifically for training the AI. Also, they could choose to Enter into agreements with organisations or individuals to use private or proprietary data, ensuring that the terms of use permit AI training.

What Does This Mean For Your Business? 

It’s possible to see both sides of the argument to a degree. For example, so many things are copyrighted and AI companies such as OpenAI with ChatGPT wouldn’t have been able to create and get a reasonable generative AI chatbot out there if it had to get consent from everyone for everything and pay for all the licenses needed.

On the other hand, it’s understandable that creatives such as artists or journalistic sources such as the New York Times are angry that their output may have been used for free (with no permission) to train an LLM and thereby create the source of its value that it may then charge users for. Although the idea of providing a way to differentiate between AI companies that had paid and acquired permission (i.e. acted ethically for their training content sounds like a fair idea), the fact that the LLMs from the main AI companies (including ChatGPT) may not even meet the criteria does make it sound a little self-defeating and potentially not that useful for the time being.

Also, some would say that relying upon companies to admit when they may have changed their AI training practices and potentially lose the certification they’ve paid for (when Fairly Trained isn’t checking anyway) may also sound as though this may not work. All that said, there are other possible alternatives (as mentioned above) that could require consent and organisations working together that could result in useful, trained LLMs without copyright headaches.

Although the Fairly Trained scheme sounds reasonable, Fairly Trained admits that it’s not a definitive answer to the problem. It’s probably more likely that the outcomes of the many lawsuits will help shape how AI companies act as regards training their LLMs in the near future.

Featured Article : NY Times Sues OpenAI And Microsoft Over Alleged Copyright

It’s been reported that The New York Times has sued OpenAI and Microsoft, alleging that they used millions of its articles without permission to help train chatbots.

The First 

It’s understood that the New York Times (NYT) is the first major US media organisation to sue ChatGPT’s creator OpenAI, plus tech giant Microsoft (which is also an OpenAI investor and creator of Copilot), over copyright issues associated with its works.

Main Allegations 

The crux of the NYT’s argument appears to be that the use of its work to create GenAI tools should come with permission and an agreement that reflects the fair value of the work. Also, it’s important in this case to note that the NYT relies on digital subscriptions rather than physical newspaper subscriptions, of which it now has 9 million+ subscribers (the relevance of which will be clear below).

With this in mind, in addition to the main allegation of training AI on its articles without permission (for free), other main allegations made by the NYT about OpenAI and Microsoft in relation to the lawsuit include :

– OpenAI and Microsoft may be trying to get a “free-ride on The Times’s massive investment in its journalism” by using it to provide another way to deliver information to readers, i.e. a way around its payment wall. For example, the NYT alleges that OpenAI and Microsoft chatbots gave users near-verbatim excerpts of its articles. The NYT’s legal team have given examples of these, such as restaurant critic Pete Wells’ 2012 review of Guy Fieri’s (of Diners, Drive-Ins, and Dives fame) “Guy’s American Kitchen & Bar”. The NYT argues that this threatens its high-quality journalism by reducing readers’ perceived need to visit its website, thereby reducing its web traffic, and potentially reducing its revenue from advertising and from the digital subscriptions that now make up most of its readership.

– Misinformation from OpenAI’s (and Microsoft’s) chatbots, in the form of errors and so-called ‘AI hallucinations’ make it harder for readers to tell fact from fiction, including when their technology falsely attributes information to the newspaper. The NYT’s legal team cite examples of where this may be the case, such as ChatGPT once falsely attributing two recommendations for office chairs to its Wirecutter product review website.

“Fair Use” And Transformative 

In their defence, Open AI and Microsoft appear likely to be relying mainly on the arguments that the training of AI on NYT’s content amounts to “fair use” and the outputs of the chatbots are “transformative.”

For example, under US law, “fair use” is a doctrine that allows limited use of copyrighted material without permission or payment, especially for purposes like criticism, comment, news reporting, teaching, scholarship, or research. Determining whether a specific use qualifies as fair use, however, will involve considering factors like the purpose and character of the usage. For example, the use must be “transformative”, i.e. adding something new or altering the original work in a significant way (often for a different purpose). OpenAI and Microsoft may therefore argue that training their AI products could potentially be seen as transformative as the AI uses the newspaper content in a way that is different from the original purpose of news reporting or commentary. However, the NYT has already stated that: “There is nothing ‘transformative’ about using The Times’s content without payment to create products that substitute for The Times and steal audiences away from it”. Any evidence of verbatim outputs may also damage the ‘transformative’ argument for OpenAI and Microsoft.

Complicated 

Although these sound like relatively clear arguments either way, there are several factors that add to the complication of this case. These include:

– The fact that OpenAI altered its products following copyright issues, thereby making it difficult to decide whether its outputs are currently enough to find liability.

– Many possible questions about the journalistic, financial, and legal implications of generative AI for news organisations.

– Broader ethical and practical dilemmas facing media companies in the age of AI.

What Is It Going To Cost? 

Given reports that talks between all three companies to avert the lawsuit have failed to resolve the matter, what the NYT wants is:

Damages of an as yet undisclosed sum, which some say could be in the $billions (given that OpenAI is valued at $80 billion and Microsoft has invested $13 billion in a for-profit subsidiary).

For OpenAI and Microsoft to destroy the chatbot models and training sets that incorporate the NYT’s material.

Many Other Examples

AI companies like OpenAI are now facing many legal challenges of a similar nature, e.g. the scraping/automatic collection of online content/data by AI without compensation, and for other related reasons. For example:

– A class action lawsuit filed in the Northern District of California accuses OpenAI and Microsoft of scraping personal data from internet users, alleging violations of privacy, intellectual property, and anti-hacking laws. The plaintiffs claim that this practice violates the Computer Fraud and Abuse Act (CFAA).

– Google has been accused in a class-action lawsuit of misusing large amounts of personal information and copyrighted material to train its AI systems. This case raises issues about the boundaries of data use and copyright infringement in the context of AI training.

– A Stability AI, Midjourney, and DeviantArt class action claims that these companies used copyrighted images to train their AI systems without permission. The key issue in this lawsuit is likely to be whether the training of AI models with copyrighted content, particularly visual art, constitutes copyright infringement. The challenge lies in proving infringement, as the generated art may not directly resemble the training images. The involvement of Large-scale Artificial Intelligence Open Network (LAION) in compiling images used for training adds another layer of complexity to the case.

– Back in February 2023, Getty Images sued Stability AI alleging that it had copied 12 million images to train its AI model without permission or compensation.

The Actors and Writers Strike 

The recent strike by Hollywood actors and writers is another example of how fears about AI, consent, and copyright, plus the possible effects of AI on eroding the value of people’s work and jeopardising their income are now of real concern. For example, the strike was primarily focused on concerns regarding the use of AI in the entertainment industry. Writers, represented by the Writers Guild of America, were worried about AI being used to write or complete scripts, potentially affecting their jobs and pay. Actors, under SAG-AFTRA, protested against proposals to use AI to scan and use their likenesses indefinitely without ongoing consent or compensation.

Disputes like this, and the many lawsuits against AI companies highlight the urgent need for clear policies and regulations on AI’s use, and the fear that AI’s advance is fast outstripping the ability for laws to keep up.

What Does This Mean For Your Business? 

We’re still very much at the beginning of a fast-evolving generative AI revolution. As such, lawsuits against AI companies like Google, Meta, Microsoft, and OpenAI are now challenging the legal limits of gathering training material for AI models from public databases. These types of cases are likely to help to shape the legal framework around what is permissible in the realm of data-scraping for AI purposes going forward.

The NYT/OpenAI/Microsoft lawsuit and other examples, therefore, demonstrate the evolving legal landscape as courts now try to grapple with the implications of AI technology on copyright, privacy, and data use laws, and its complexities. Each case will contribute to defining the boundaries and acceptable practices in the use of online content for AI training purposes, and it will be very interesting to see whether arguments like “fair use” are enough to stand up to the pressure from multiple companies and industries. It will also be interesting to see what penalties (if things go the wrong way for OpenAI and others) will be deemed suitable, both in terms of possible compensation and/or the destruction of whole models and training sets.

For businesses (who are now able to create their own specialised, tailored chatbots), these major lawsuits should serve as a warning to be very careful in the training of their chatbots and to think carefully about any legal implications, and to focus on creating chatbots that are not just effective but are also likely to be compliant.