The AI Issue 🤖 Platform AI disclosures are here, LAION fights back against photographers, EU moves ahead, A chat with the Data Diva, and much more!
This is Creator Economy Law, a newsletter dedicated to exploring and analyzing the legal issues surrounding the creator economy, creators, and internet platforms. If you enjoy what you’re reading, share with friends, and invite them to subscribe using the button above and share using #CreatorEconomyLaw
The first Monday in May has come and gone, but we’re still talking about the Met Gala! The theme this year was “Karl Lagerfeld: A Line of Beauty” in honor of the late designer known for revitalizing the Chanel brand. Lagerfeld also had a cat named Choupette, which served as the theme for many outfit choices. Speaking of… Vogue has the complete rundown of all looks from the evening!
Here’s what’s been happening in the world of Creator Economy Law.
AI disclosure requirements have arrived… sorta.
As platforms continue adopting artificial intelligence tools and providing those to end users to assist with content creation, the question is coming up as to what disclosures, if any, are necessary or legally required to accompany such content. I’ve linked below under the “Read” section some examples of early policy adoption that is taking place across some platforms, currently in the absence of any explicit regulation on the topic.
And for companies that are deploying AI tools within their products and services, the Federal Trade Commission has yet another warning.
“If we haven’t made it obvious yet, FTC staff is focusing intensely on how companies may choose to use AI technology, including new generative AI tools, in ways that can have actual and substantial impact on consumers.”-FTC
Add to the mix Chair Lina Khan’s opinion piece in The New York Times that highlights a few key issues with AI: (1) large tech corporation dominance being further solidified; (2) control of raw materials needed for AI tool development, including processors, data, and cloud technology; and (3) the promotion of fair competition and protection of consumers from unfair and deceptive practices through the deployment of AI tech.
On Thursday, the White House is hosting leaders from Google, Microsoft, OpenAI, and Anthropic to meet with Vice President Kamala Harris and senior Administration officials.
LinkedIn is reportedly piloting a new generative AI-powered tool that will provide job seekers with generated messages they can send to hiring managers. TikTok is exploring official guidance and tools for disclosure. Meta‘s Mark Zuckerberg teased AI tools that are in the works for integration across all its platforms, from social to messaging.
Meanwhile, creators can consider adding their own disclosure as part of their content. For example, an opening card to a video such as this one from an AI-generated version of the Britney Spears track ‘Circus’ re-imagined as a duet with Michael Jackson (watch on YouTube unless it gets taken down).
Does a disclosure in this instance change anything, legally speaking? Probably not, even under a First Amendment argument. Also, this particular AI-generated track is heavily reliant upon (1) music publishing controlled by major music publishers; and (2) a significant portion of the original sound recording.
- TikTok – “Synthetic and Manipulated Media”; Also, The Information‘s Kaya Yurieff reports that TikTok is working on a disclosure tool.
- LinkedIn – “Best practices for content created with the help of AI” (H/T to Kaya again!)
- Facebook and Instagram – The company’s Misinformation policy covers manipulated media, which includes AI/ML and other synthetic media. It might be a stretch, but the platform’s Inauthentic Behavior policy includes as part of the definition of “inauthentic behavior” the use of the platform to “mislead people or Facebook […] About the source or origin of content.”
- Twitter – Whether or not it’s still accurate, in 2020 a blog post was published “Building rules in public: Our approach to synthetic & manipulated media”
🗣 Franklin’s Take: Here are three options I came up (and shared two months ago) with that I hope we start seeing widely adopted and implemented:
- A self-disclosure toggle. Much like YouTube offers a “Contains paid endorsement” option for creators upon upload, a toggle that is part of the UX and UI that allows for a means built into the product is important. I wouldn’t be surprised if regulations address a disclosure obligation in the future.
- Copy-and-paste detection. Social media sites already track our every move, including keyboard inputs and clipboard data access. This can be a means of detecting whether or not a user is drafting a post over a period of time, with one or more versions. Or, did they paste a block of text and then hit post? This detection method is not perfect, since I could easily be manually typing this from a text generator and into the post creation tool on a platform. Or, I could have saved a post that I drafted in my Notes app (as I often do), waiting to then paste it into a post instead of using a scheduler. And, third-party scheduling tools pose a problem here, too, since a platform would be relying on data captured by that platform, if any, at the time of composing.
- Automatic disclosures. This would apply to platforms that offer generative technologies to their end users to assist (or replace) content creation. There’s also an interesting #Section230 liability twist here if the content is no longer user-generated content, but rather AI-generated content from a tool the platform (or its partners) provides. Does the platform lose that safe harbor?
LAION Fights Back Against Photographers with Copyright Law
LAION is reportedly using German copyright law to fight back against photographers that are sending takedown requests. The non-profit research group is using the law firm Heidrich Rechtsanwälte to send responses to people sending in takedowns, according to a report by ProfiFoto (which I translated from German using Safari).
The responses are reportedly highlighting laws that allow LAION to demand damages for the sending of false claims.
“We also point out that our client can assert claims for damages in accordance with § 97a para. 4 UrhG if they are unjustified in terms of copyright,” the firm purportedly wrote in a response to Robert Kneschke, a German stock photographer.
For its part, LAION has been clear for quite a while on its website about the nuances of its database.
LAION datasets are simply indexes to the internet, i.e. lists of URLs to the original images together with the ALT texts found linked to those images. While we downloaded and calculated CLIP embeddings of the pictures to compute similarity scores between pictures and texts, we subsequently discarded all the photos. Any researcher using the datasets must reconstruct the images data by downloading the subset they are interested in.
Following what appears to be an escalation of multiple letters back and forth (isn’t lawyering thrilling 🤣) a recent letter from LAION’s counsel includes a demand for damages, as permitted by statute, for €900.
But things get stickier…
Trevor Baylis on Twitter pointed out the interesting provisions in § 60d para. 4 UrhG that seem to indicate these broad releases and commercial uses of the LAION dataset go against German law.
📖 Read: “Laion droht Kneschke” via ProfiFoto.
🗣 Franklin’s Take: I’m curious as to whether or not there’s any form of reasonableness or another standard that would allow for a photographer’s warning/demand under § 97a to not be deemed “unjustified” or “ineffective”.
Also, Sylvie Fodor pointed out in a comment on my post earlier in the week an interesting nuance to this: “The mentioned provisions in the German Copyright Law are transposition of Art. 4 of the EU copyright directive on Text and Data Mining exceptions. This means that this lawsuit will be a test of the said article and has meaning across the EU.”
The EU’s AI Act is moving ahead
The European Parliament has reached an agreement on a finalized draft of the AI Act, moving forward in the process of it becoming law. Read more from Luca Bertuzzi on EURACTIV. Here are two takeaways:
🤖 Generative AI tools (referred to as “general purpose AI systems”) will have to disclose what, if any, copyrighted materials were used for training; and
🤖 Three risk levels remain: prohibited, high, and low.
There’s a lot to unpack, especially the freedom of expression elements, but the #copyright scope is the most recent development. The next step now is when EU member states start to discuss and work towards finalized a version.
Read the most recently published draft, dated December 6, 2022.
Stability AI seeks to dismiss Getty lawsuit
On May 2, Stability AI filed a couple of documents, including its motion to dismiss the lawsuit brought by Getty Images.
The focus of the argument: Stability AI is a UK-based company, and, therefore, Getty Images doesn’t have standing to bring the lawsuit in Delaware courts.
Also, they argue the training of the Stable Diffusion AI models took place in the UK, outside of Delaware, and that activity is the basis for the allegations of infringement being made by Getty.
Lastly, Stability AI argues the case could be transferred to be included as part of the class action brought against Stability in California (the Anderson case).
Here are some takeaways I put together:
🤖 This is largely a procedural and technical move to get the case dismissed, not necessarily anything specific to copyright law (yet).
🤖 It’s interesting that Stability AI is using the UK jurisdiction as the shield for the training activity. Remember, the UK has the text and data mining (TDM) exception to copyright infringement. So, this could be the start of seeing how that holds up: (1) across jurisdictions; and (2) outside of the research and academic environments in which the TDM exception only applies.
🤖 If Getty is added to the class as a named party, that could be very interesting given their claims arguably would support the class moving forward with specific, substantially similar outputs that could be shown to sustain the copyright infringement claims on the outputs, plus the registration issues.
Read the full motion to dismiss, and other recent filings, over on CourtListener.
Copyright Office Listening Sessions Underway
During this week’s U.S. Copyright Office AI Listening Session for Visual Arts, Heather Whitney mentioned Kristina Kashtanova’s ‘Rose Enigma’ registration. It’s an extremely interesting next step in AI for the USCO to consider!
In March, Kashtanova submitted a cover letter as part of their registration application for ‘Rose Enigma’.
The final work was created using a combination of tools that started with both: (1) crafting text prompts, based in part on how the software would interpret the text inputs; and (2) a hand drawn depiction on pen and paper of the cyborg. The text prompts and the drawing were both used as inputs.
Additionally, Kashtanova explains their use of additional controls within the Stable Diffusion software that contributed to their ability to exercise control over the creative process, as well as the use of the ControlNet Depth model to help further control Stable Diffusion.
Shout out to Joe Gratz who’s also signed on to the cover letter!
It’s not too late to register to attend the final two listening sessions. Visit copyright.gov/ai/ or use the links below:
- May 17, 2023 – Audiovisual Works Listening Session
- May 31, 2023 – Music and Sound Recordings Listening Session
Generative AI Thoughts. Josh Marshall, founder and Editor-in-chief of TPM, published a thought piece on generative AI, copyright, data scraping, and more. It’s definitely worth a read for all creators and people that work with creators.
Buzzfeed Wants You. In an interview with Axios‘s Sara Fischer, BuzzFeed CEO Jonah Peretti is looking to maximize content creation by working with creators through what’s being deemed a “residency” program. My question for creators is: Do you want Buzzfeed?
Peretti also published “5 Predictions for the Future of Digital Media”
Controversy Still Sells. Sanya Burgess explores how the recent arrest and Twitter reinstatement of controversial UK social media figure Andrew Tate may have actually helped boost advertising revenues for the platform, estimated to be hovering at £10m a year. Check out the article (which includes an option to listen) on Sky News.
Follow the money, and your gut. Tubefilter‘s James Hale reports on a recent talent shakeup in the VTuber space following the departure of three top Vtubers from VShojo. In an April 26 tweet thread, Veibae said she chose to leave virtual talent agency VShojo because “new contracts didn’t make any financial sense to me.”
Lossy Audio. Larry Mills over at Pex published an insightful piece highlighting the issues with platforms that offer original audio type editing tools. Check out: The state of modified audio: Social media trends are diverting royalties away from rightsholders
On Tuesday, the National Center for Missing & Exploited Children published the 2022 CyberTipline Reports by Electronic Service Providers (ESP). As covered by Cecilia D’Anastasio for Bloomberg, some of the biggest platforms reported an increase in reports.
Former Twitter owner and co-founder Jack Dorsey spent some time posting on Blue Sky (his recently-launched Twitter competitor) to criticize Elon Musk‘s recent operations (or lack thereof?) of the Twitter platform. (Ashley Capoot via CNBC) Dorsey has a point, as the recent unpredictability of the platform has brands concerned and less likely to invest ad spend into the platform.
“Verified gov or publicly owned services who tweet weather alerts, transport updates and emergency notifications may use the API, for these critical purposes, for free,” the company announced in a tweet from the TwitterDev account. The move is yet another backtracking on the recent high-cost API developer access plans put in place.
Elon tweeted that soon (he said “next month” on April 29th, so did he mean May or June?) media publishers will be able “to charge users on a per article basis with one click.” It’s meant to offer an alternative to users that may not subscribe for an entire month or year to a media outlet. He called it a “win-win”, but historically it never seems like a great thing for publishers.
The Federal Trade Commission is claiming Facebook violated its prior 2020 privacy order and now proposes a blanket prohibition that would prevent the platform from profiting from data it collects, including through its virtual reality products, from users under the age of 18. It would also be subject to other expanded limitations, including in its use of facial recognition technology, and required to provide additional protections for users. Read the full announcement.
Meta is moving away from its original content production, opting to close down its Facebook Watch operations and let go of its leader Mina Lefevre in the recent round of layoffs. (Sam Gutelle via Tubefilter)
Meta is working overtime to combat hackers and other bad actors. (Lily Hay Newman via WIRED) Also, check out the company’s press release on the efforts to protect businesses, and a recent blog post from the Engineering at Meta team that explores the malware threat landscape.
Instagram is offering brand users with a shop account the ability to take user-generated content that has tagged the brand and associate it with the product on the product page and storefronts. Most importantly, the feature requires the brand to request permission from the content creator, who can then accept or reject the request. Shout out to 🌠 Yolandé Haynes for posting about this on LinkedIn.
Insta is also renaming branded content ads to partnership ads and allowing advertisers to amplify content from a creator or other partner’s handle, to scale their collaborations. Read the official announcement.
Don’t miss a new investigative piece by Katie McQue and Mei-Ling McNamara, published by The Guardian, that was 2 years in the making and that explores alleged issues Meta faces with controlling child sex trafficking activities on its platform.
The EU-U.S. data transfers are an ongoing concern for Meta, as they detailed in recent SEC filings. Joe Duball published an insightful article on the IAPP – International Association of Privacy Professionals website.
ByteDance is working on an AI music app, according to Music Business Worldwide (MBW)‘s review of recent job openings. Trademark filings uncovered by Business Insider also allude to TikTok‘s exploration of launching a book publishing arm, called 8th Note Press. (Read on Apple News+)
According to Reuters,Senate Majority Leader Chuck Schumer said during a press conference on Wednesday that senators will start looking into the legislation to address a ban on TikTok and other foreign-owned apps that pose a threat to national security.
TikTok also had a presence at IAB’s NewFronts, where they announced TikTok Pulse, a tool in which content publishers can sell ads alongside their own content as part of a 50/50 revenue split. In a new Ads support article, the company notes Pulse “enables advertisers to align their brand with the videos driving TikTok culture by showing their ads immediately after the top 4% content on the platform within a certain category (for example, Beauty Pulse, Sports Pulse, etc).”
TikTok announced an updated Creativity Program Beta to U.S. creators that meet eligibility requirements, but on an invite-only basis for the next couple of months. The program is separate from the Creator Fund, and creators in that program can switch (but can’t switch back). The eligibility requirements include:
- At least 18 years of age
- 10,000 follows
- 100,000 authentic video views in the last 30 days
Want a free Apple AirTag? The NYPD is giving them away for free in response to the KiaBoyz challenge on TikTok that’s led to a string of car thefts. It may not work for long, though, given the recent announcement that Apple and Google are working together to prevent unwanted tracking with products like the AirTag.
Fortune writes, “YouTube’s pitch was very clear: If you want to reach Gen Z—consumers born between 1997 and 2012—YouTube is the place to be.” I highly recommend you check out Alexandra Sternlicht‘s recap of the event. (Read on Apple News+)https://www.linkedin.com/embeds/publishingEmbed.html?articleId=9125139934589397399&li_theme=light
YouTube announced, via the Creator Insider channel, a policy update to its YouTube Partner Program and the re-application process.
Snap presented at IAB NewFronts on Tuesday and announced several new ad formats, content partnerships, and ways to work with creators. Snap mentions they have 750 million monthly active users.
Utah gets challenged. The Free Speech Coalition has filed a lawsuit against Utah’s age-verification law, they announced in a press release. The complaint asks for declaratory and injunctive relief against the law on account of violations of the 1st and 14th Amendments. Joining Free Speech Coalition in filing the challenge are Andrea Barrica, founder of the sex education site O.school; journalist, educator, and content creator Charyn “Ryn” Pfeuffer; Utah-based erotica writer D.S. Dawson; “John Doe,” a pseudonymous Utah-based attorney; and fan platform JustFor.Fans. The parties are represented by Jeffrey Sandman of Webb Daniel Friedlander LLP, D. Gill Sperlein of the Law Office of D. Gill Sperlein, and Jerome Mooney of Weston, Garrou & Mooney Law Firm on a pro bono basis. The case is Free Speech Coalition v. Anderson (2:23-cv-00287), District Court, D. Utah.
Protect the Children! Laws aiming to protect children online will have an impact on how adults use and enjoy online activities, soon with less anonymity through required verification of their real identities. Natasha Singer writes about the issue in The New York Times.
Taking the good with the bad. A new McKinsey Health Institute survey finds that Gen Z’s social media engagement can feel negative but can also help with finding mental health support and connectivity. Check out the full publication.
Live Shopping in China. Despite experiencing growing pains here in the U.S., the livestream shopping craze is strong in other parts of the world, like China. “Last year alone, an estimated $500 billion in goods were sold via livestream on apps like Douyin, the Chinese version of TikTok, or Kuaishou, another short video platform — an eightfold increase since 2019.” Check out the in-depth piece by Vivian Wang in The New York Times.
Speaking of shopping. Cordilia James and Julie Jargon break down some of the latest product scams taking place across social media platforms, the Federal Trade Commission‘s efforts to enforce consumer protection laws, and how users can protect themselves when shopping online via social media ads. Read the full piece over at The Wall Street Journal.
Can’t afford that NFT? No worries. The marketplace Blur launched a new lending option called Blur Lending, or Blend. “With Blend,” they explained on Twitter, “NFT holders can borrow ETH against their NFTs without needing to sell.” Or, if someone can’t afford an NFT, like a CryptoPunk, the platform offers a buy-now-pay-later system. Read more from Cam Thompson over on Coinbase.
The Lawfare Institute Senior Editor Quinta Jurecic and Matt Perault, Director of the Center on Technology and Policy at UNC-Chapel Hill, talk about Section 230 and its application to generative AI with the original drafters of Section 230, Senator Ron Wyden and Chris Cox, formerly a U.S. congressman and SEC chairman. (Spoiler: they don’t think it applies!) Shout out to Jess Miers who gets a mention during the episode with the alternate view that Section 230 should apply.
Want to get caught up on FAST? I’ve written about free ad-supported streaming television (FAST), as well as advertising-based video on demand (AVOD) in previous issues, but there’s an excellent overview of the streaming advertising ecosystem by David Pierce over on The Verge.
You’re gonna hear me roar. 🦁 In a recent decision out of Australia, singer Katy Perry lost a #trademark dispute with designer Katie Perry over the use of the name on clothing labels. “This is a tale of two women, two teenage dreams and one name,” the judge wrote
The lawsuit has been on-going since 2019 😱 It targeted Katy Perry’s sale of merch during her 2014 and 2018 tours. But, the two have been at odds since as early as 2008 when Katie Perry (now Katie Taylor) first filed to register the trademark in Australia. #TrademarkLaw attorneys: let me know what you think about this decision in the comments!
Going once… VICE Media is reportedly soon to be entering into bankruptcy after failing to find a buyer. It could be only a matter of days according to reporting by Lauren Hirsch and Benjamin Mullin for The New York Times. (See also from Reuters)
Wikipedia to Avoid UK Age Checks. The Wikimedia Foundation, which runs the online, community-powered encyclopedia Wikipedia, is pushing back on the UK’s pending legislation The Online Safety Bill. The bill is set to impose obligations on tech platforms “to protect users from harmful or illegal content and is expected to come fully into force some time in 2024.” (Christopher Vallance & Tom Gerken via BBC News)
Brazil has been busy. The country’s Fake News Law (Bill 2630) is receiving heavy pushback from tech companies. Also, Telegram was handed a win by a judge that lifted a previous suspension order for the app across the entire country, noting it wasn’t reasonable to impact all users given the ongoing investigation.
It’s okay to cry sometimes. And that might just be what UK retailer Boohoo is doing after reaching a $197 million settlement agreement for a class action lawsuit brought in California over its false pricing and sale tactics. Robert Freund provides an awesome breakdown on Twitter, but also landed a feature in the Business Insider coverage by Jordan Hart. (Read on Apple News+)
The Data Diva E161 – Vivek Kumar and Debbie Reynolds – "The Data Diva" Talks Privacy Podcast
Thanks to Debbie Reynolds for having me on the latest episode of her podcast, “The Data Diva” Talks Privacy Podcast! Our discussion hits on the crazy wide range of both my passion projects and my professional work.
It was fun to chat about:
😎 The “why” behind starting my newsletter on #CreatorEconomy
🤹♂️ My career journey and how all of my passions have managed to intersect at every stage of my professional life
🕵️ Privacy, children, and liability within the creator economy
🤖 The hot topic of the year: artificial intelligence and intellectual property
🤩 Rights of privacy and publicity
Generative AI Meets Copyright
I recommend watching Pamela Samuelson‘s recent talk as part of the CITRIS Research Exchange that was co-hosted with the UC Berkeley Artificial Intelligence Research Group (BAIR). It offers a great overview of current copyright issues with generative AI. I also appreciate hearing from a wide range of perspectives on the topic.
Are you learning anything new this week? Let me know in the comments! I’d love to check it out.
This week, Ed Sheeran testified in court as part of a copyright infringement lawsuit that was brought by the heirs of Ed Townsend, a co-writer with Marvin Gaye on the track ‘Let’s Get It On‘. NBC News’s Jacob Soboroff, reporting for TODAY, gives a great rundown of the case, and a comparison of the songs. It was left up to a jury to decide whether or not Sheeran infringed on Gaye’s work, and Bloomberg Law announced he was found not guilty of copying!
Affiliate Links. As an Amazon Associate, I earn from qualifying purchases. I have noted above where links to products on Amazon may earn me a commission if you make a purchase. Thanks for supporting my work!
Not Legal Advice. This newsletter is published solely for educational and entertainment value. Nothing in this newsletter should be considered legal advice. If you need legal assistance or have specific questions, you should consult a licensed attorney in your jurisdiction. I am not your attorney. Do not share any information in the comments you should keep confidential.
Personal Opinions. The opinions and thoughts shared in this newsletter are my own, and not those of my employer or any of the third parties mentioned or linked to in this newsletter. No affiliation or endorsement is implied or otherwise intended with third parties that are referenced or linked.