Oliver Scherenberg

Lethe GmbH

CFO and Chief Legal and Compliance Officer

VON HAVE FEY Rechtsanwälte

Of Counsel (attorney in a passive role)

SCHERENBERG LEGAL & LICENSING

German law firm with a focus on the monetization of intellectual property rights, web3 and generative AI.

Founder and IP Licensing Counsel

WEB3LEX

Award-winning international online knowledge base for web3 and AI legal and tax topics and home of the international #lawyers4web3 network. 

Browse our knowledge base with an AI assistant for the legal answers you are looking for in the areas of web3 and AI (currently testing, published soon).

WEB3LEX is a virtual law firm, which puts you in touch with web3 legal and tax experts from over 60 jurisdictions. Book your meeting through the platform and pre-pay with credit card or PayPal (currently running a pilot on this feature). 

Want to have an immersive meeting with your WEB3LEX lawyer? No problem, come visit our Virtual Office with your fancy avatar.

- Winner of 1 golden and 3 bronze nails at the 2023 creative festival of Art Directors Club.
- Winner of 4 Red Dot Awards (among one "Best of the Best") 2023.
- Short-listed for European Metaverse Awards 2023.

Founder

DRYMSTRYM TV GmbH

Multi-Channel-Network with a German broadcasting license. Are you running a live streaming or SmartTV channel and require a license? We can legalize your channel by granting you a sub-license (approved by the media regulators).

Co-Founder and Managing Director

The Orthocorp Ltd.

web3 voxel game and Metaverse. An innovative experiment on the Ethereum Blockchain. Doomed, but what isn't?

Orthosigliere (The Legal Guy)

Sisvel Group

General Counsel Legal and Business

Preu Bohlig & Partner

Partner

Unverzagt von Have

Associate Partner

Freshfields Bruckhaus Deringer

Associate

Licensing, web3 and Media Law.

Picture courtesy of Rainer Hosch (www.rainerhosch.com), 2022.

Titanic II in the making – for younger viewers.


If the ship were to sink today, this might be the final scene of the movie.

Considering today's social media habits, the sad truth is: this probably wouldn’t even be unrealistic.

And yes, I know: the ship didn’t actually sink stern-first but went down in a bow-first nosedive.

Maybe AI knows something we don’t!?

Folks, we might be onto something here...

#conspiracy #bestselfie #titanic2 #howhavewegottenhere

German Court: AI trainig can be considered copyright infringement

The Munich Regional Court delivered an interesting decision today.

GEMA, the collective society for music rights holders, sued #OpenAI over lyrics used to train and reproduced by ChatGPT.

And GEMA fully won. Here are the key points of the judgement:

1️⃣ OUTPUT infringement:
ChatGPT reproduced lyrics of several songs verbatim. The court saw that as a copyright violation.
👉 No surprise here. If the output looks like the original work, that’s an infringement.

2️⃣ INPUT (= training data) infringement:
Because the lyrics appeared word-for-word, the court concluded they must have been part of the training data. That means, there must be a “copy” of the work in the training data.
👉 I believe this is technically wrong. AI models do not store original data. They learn patterns. This point may not survive appeal.

3️⃣ No text and data mining (TDM) privilege:
Since the court found that data was stored, they come to the conclusion that this does not fall under the TDM privilege.
👉 I am afraid this is also incorrect, since no data is stored in AI models to begin with.

4️⃣ Liability:
OpenAI, not individual users, was found responsible for near-verbatim reproductions.
👉 That makes sense under the Court's logic. But it does not stop there. Make no mistake: users are not automatically off the hook. Just like in pre-AI times: If you publish infringing material, you are still responsible, regardless of the source.

GEMA is the clear winner of round one. Now both sides head back to their corners. OpenAI will spit out some blood, and get ready for round two. This ain't over yet.

Ding ding!🥊

See addtional posts and many more details and perspectives regaring this case by my esteemed colleagues Dr. Ralph Oliver Graef, LL.M. (NYU), Dr. Barry Scannell, Max-Julian Wiedemann, LL.M., Edward Lee and many others.

Preparing early for the upcoming corporate Christmas party this year. 



P.S.: I made this picture last year already with the then most recent Midjourney AI diffusion model. 

P.P.S.: I seriously consider getting this made by some talented knitting expert on Etsy or the like. Recommendations anyone?

No alternative text description for this image

This morning we have obtained a landmark court decison regarding the use of copyright protected works for the training and operation of AI models under EU law. 🏆

In its ruling in GEMA vs. OpenAI the Munich Regional Court found that OpenAI should have asked for permission before feeding our members’ song lyrics into ChatGPT. 

According to the court AI training with copyrighted works and the memorization of such works in the model without prior authorisation constitutes a violation of copyright rules  - and triggers AI companies’ liability. 

This is a huge success for all whose creative output has been used to build and train an AI model. Creators’ rights are not up for grabs. 

We will continue our mission to defend our members’ rights! Next stop: GEMA vs. SunoAI in January.  
#fairness #AI #copyright

taplio