Longer online version: print version here

Many lawsuits face the internet giants

PROBABLY the biggest copyright news now is the Google search engine continuing its war on law - determined to evade any legal requirement to pay for the news from which it makes profits. Much else is happening:

From whom does ‘artificial intelligence’ learn?

So-called "artificial intelligence" company OpenAI lists "books" among its "assets". Were these scraped from a legal source? Publishers and authors will want to know. Or do they come from schemes of dubious legality such as those of the Internet Archive, which publishers are suing? [Update below.]

Do not cross the translator

Translator Yilin Wang expressed loud and effective anger at the British Museum using her translations of the poet Qui Jin without permission, credit or payment. On 23 June the Museum offered to pay for use up to the time when it removed them.

10 July 2023 Yilin Wang has engaged London lawyers Howard Kennedy LLP to sue the British Museum, reports The Art Newspaper. At the time of writing she had crowdfunded over £17,000 towards costs. She states that she intends to go to the Intellectual Property Enterprise Court (the copyright small claims track) for breach of copyright and breach of her moral right to be identified as the translator. There are some interesting legal questions: for example Yilin Wang is based in Vancouver and we find no mention in Canadian copyright law of a requirement to "assert" the moral right to be identified before enforcing it - a quirk of UK law. The Freelance is looking into these.

Mark Stephens, a partner at Howard Kennedy LLP, says the British Museum has "handled this in a high-handed and cavalier manner... It is really shoddy for a large, national organisation that holds itself out as a bastion of British enlightenment and fair play to conduct themselves in this way... [this is] just the latest example of the British Museum's colonialist attitudes and rapacious spoliation of other cultures."

Twitter in yet more trouble

At the time of writing the social media network Twitter is notoriously in trouble. Limiting the number of posts users can read is quite remarkable for a network that relies on eyeballs bouncing off adverts for revenue. Rather less noisily, music publishers announced on 15 June that they are suing Twitter. The 17 publishers want Twitter to pay for a licence to cover user uploads of copyright music - as do TikTok, Facebook, Instagram, YouTube and Snapchat. They want $250 million for specific uploads of 1700 songs. They bring in evidence Tweets by proprietor Elon Musk claiming that copyright "goes absurdly far beyond protecting the original creator" and that "overzealous" application of copyright laws "is a plague on humanity". Those could demonstrate flagrant disregard for the law. Under US law the publishers can claim $150,000 for each infringement of a registered work.

First ChatGPT defamation lawsuit falls

Not strictly, yet, a copyright matter, but: lawyers for the Australian mayor who threatened proceedings against ChatGPT for making up a libellous biography for him told journalist Ashley Bellanger that "no further legal steps are being taken in this case at this time after OpenAI removed the defamatory information and 'corrected the public record'." The Freelance is not clear that it is possible to remove such a thing from a Large Language Model. More work required.

7 July 2023

Novelists sue

Authors Mona Awad and Paul Tremblay are suing OpenAI, owners of ChatGPT, for breach of copyright. They say the text generator produced accurate summaries of their work. They are doing so in Federal court in San Francisco.

Under UK law, Awad and Tremblay would have to show that the system copied a "substantial part" of their work, though they could argue that it copied their work into the system. US law is often more amenable to protecting the gist of a work or characters in it, but also suffers from the so-called "fair use" doctrine, which leaves it largely to each court to decide whether copying is permitted as an "exception" to copyright. The case is a "class action" suit under US law, which means others may be able to join.

The lawsuit refers to the mysterious "books asset" mentioned above.

Privacy, and apocalypse too...

Another lawsuit, also in Federal court in San Francisco, alleges that both ChatGPT and the image generator Dall-E infringed the privacy rights of 16 anonymous claimants and others who may join the "class". It refers both to privacy interests in material "scraped" from the Web, and in data collected from those who sign up to use or explore the systems, which would include the present author on both counts.

The Freelance does not pretend yet to have read all the 160-page complaint, but in addition to privacy it refers to general threats of "AI", including "AI-Fueled Misinformation Campaigns, Targeted Attacks, Sex Crimes, and Bias; Hypercharged Malware Creation; Autonomous Weapons"; and for good measure "Larceny/Receipt of Stolen Property".

10 July 2023

Yet more authors sue OpenAI and ChatGPT - and Facebook

The latest case for breach of copyright by ChatGPT (of which we are aware) involves comedian and author Sarah Silverman and authors Christopher Golden and Richard Kadrey. The three are suing OpenAI, owner of ChatGPT, and Meta, owner of Facebook. The complaint is filed by the same lawyers at that of Mona Awad and Paul Tremblay (above) in the same court. The case against Facebook concerns the sources of training data for its own suite of "AI" tools.

11 July 2023

Facebook in deep trouble

Again, this is not strictly a copyright matter - but the fate of arguably one of the largest infringers is important. On 4 July the Court of Justice of the European Union (CJEU) issued a ruling on a German decision against Facebook under competition law.

The job of the CJEU was to decide whether a ruling by the German competition authority (the Bundeskartellamt or Federal Cartel Office) was compliant with EU law. Specifically: could the German authority take privacy law (such as the General Data Protection Regulation, the GDPR) into account when deciding that Meta Platforms Inc (formerly Facebook Inc) was abusing a dominant position - acting anti-competitively, as a monopoly does?

Yes, says the CJEU. And users must give explicit consent before Facebook collects their data - including data collected through mini-apps or "widgets" on sites other than facebook.com, which may account for the majority of Meta's hoard of personal information. (This is why there are no links to Facebook from londonfreelance.org, there are no widgets here and we set no cookies.) The existence of a dominant position is relevant to deciding whether that consent is freely given.

The Cartel Office ruling that Meta Inc cannot share personal data between its different services - including Facebook, WhatsApp, Instagram and now Threads - is consistent with EU law. It is referred back to the German court that sought that guidance from the CJEU. That's Meta's business model broken, as the CJEU implicitly acknowleges. And that's a very good reason for its new Threads service not being available in the EU.

Google wages war on law can pay, won't pay for news

Squid Game creator lost the contest signed his rights away