Director of policy at Open Future. Advocate for more open, just & inclusive digital policies in Europe & beyond. Collector of collateral knowledge. Depending on the task I can shape-shift between being a systems architect, an analysis, an activist or a cyclist.
Last week, the third draft of the Code of Practice for General Purpose #AI was released. It contains a rather curious limitation, as @paulk points out on the COMMUNIA blog. Is web scraping really the only concern for AI?
Read more below ⤵️
https://communia-association.org/2025/03/21/is-web-scraping-the-only-copyright-concern-for-ai-the-code-of-practices-blind-spot/
We have just launched the website for the CommonsDB project where we are documenting our work on building a public registry for #PublicDomain and openly licensed works:
If you are interested in #copyright registries & want to follow our work you can subscribe to our newsletter.
🚀 The first #CommonsDB newsletter features insights from @paulk and Sebastian Posth on building a public registry for Public Domain and openly licensed works.
📬 Don’t miss out—subscribe to get the next issue delivered straight to your inbox: https://www.commonsdb.org/blog/commonsdb-a-foundation-for-transparent-digital-rights-information/?ref=commonsdb-newsletter
Our latest report explores 20 years of EU policies supporting Digital Commons, exploring the shift from Open Access to collective governance of key digital infrastructures.
🔗 Read more here: https://openfuture.eu/publication/from-open-access-to-collective-governance/
As the 🇪🇺redefines its industrial strategy amidst rising geopolitical tensions and external interference, the report highlights how commons-based approaches can support innovation and autonomy.
In a recent talk at the European Heritage Hub Forum, @paulk talked about the implications of #AI systems for the role of cultural heritage institutions today.
Watch the video: https://openfuture.eu/blog/museums-and-ai-balancing-innovation-and-integrity/
looks like the #ai-copycalipse is being delayed. @BGH_Bund rules that flying (killer) robots cannot rely on the freedom of panorama exception (or so): https://bundesgerichtshof.de/SharedDocs/Pressemitteilungen/DE/2024/2024204.html?nn=10690868
Machine readable or not? Some (second ✋) observations on the hearing in the case Kneschke vs Laion e.v. This is the first case about the use of a copyrighted work as training data for generative #AI models, and unsurprisingly centers on the question what exactly constitutes a machine readable opt out under article 4 of the #copyright directive https://copyrightblog.kluweriplaw.com/2024/07/22/machine-readable-or-not-notes-on-the-hearing-in-laion-e-v-vs-kneschke/
New paper from @shayneredford et al showing abrupt rise in restrictions on access to online content via robots.txt & ToS over the last year due to the sudden realization that content is scraped to train generative AI systems. This shows a clear need to come up with a standardized vocabulary to adress data collection for different use cases to make it possible to opt-out of generative AI training while allowing other uses: https://www.dataprovenance.org/consent-in-crisis-paper
This is a perfectly reasonable observation and i do not see any reason why this should require anonymity. More importantly tough granting a someone anonymity due to “an evolving situation” is a pretty low journalistic standard on behalf of @euractiv (in the end all situations are evolving)
🏠🔙🧠! - from the European Peoples Party “5-point plan for a strong Europe that protects and shapes the future”
Dit was de PublicSpaces Conferentie 2024! Twee dagen lang hebben we gesproken, nagedacht, en gediscussieerd over het internet van morgen - een internet waar we zelf de regels bepalen!
Bedankt voor het mee-creëren van een hoopgevende en inspirerende conferentie.
Take back the internet!
Volgend jaar zijn we terug! Save the date:
PublicSpaces conferentie 2025: 12 en 13 juni
De opnames van 3 zalen kun je binnenkort terugkijken op PeerTube: https://video.publicspaces.net/
#pubconf2024
Today marks 5 years since the EU's #copyright directive came into force. It is clear that the #article17 #uploadfilters did not kill the internet, which raises the question of whether our fight to #savetheinterent was worth it.
In this post, I argue that it was very well worth the collective effort and that our activism changed many thinks for the better: https://copyrightblog.kluweriplaw.com/2024/06/07/article-17-five-years-later/
Aujourd'hui, nous sommes aux @NGICommons pour construire collectivement une stratégie pour diffuser & renforcer la place des #communsnumériques dans les politiques publiques européennes.
🌐Une démarche à laquelle nous sommes fiers de prendre part ! Merci pour l'invitation🙏
@moniquecalisti @OpenForumEurope @jankrewer @paulk @openfuture
@jeffjarvis short summary here: https://re-publica.com/en/node/6013
wrapping up day 1 of #republica24 with a 🔥side chat between @leonido and Peter Baldwin on the paradox of the #publicdomain in the digital age.
"What is needed, instead of efforts to undermine the existing framework, are measures to ensure that the current approach can work in practice" — @paulk's thoughts on Poland challenging the rule of EU #copyright law because of #AI: https://copyrightblog.kluweriplaw.com/2024/02/20/tdm-poland-challenges-the-rule-of-eu-copyright-law/
When you “de-bias” your AI model and your AI model really only is a stochastic next pixel predictor that has no understanding of the world: https://www.theverge.com/2024/2/21/24079371/google-ai-gemini-generative-inaccurate-historical
@magnusmanske @mathias thank you. i assume that counter is for *files* and looking at the examples it seems that many of them are different versions (sizes) of the same work. I guess that means that the actual number of works is significantly lower (if the first 200 pics are a representative sample the actual number of works would be 60% of the number of files so more like ~6M works.
Now the text of the #AIAct is available (🙏🏼@bertuzluca), it's time to see how my analysis of the © provisions from December holds up: Nothing of substance has changed. There is a new recital 60ka requiring the AI Office not to assess © compliance work-by-work.
This is welcome, though with datasets with billions of works, such assessments are not feasible anyway. This shows that the EU must now build infrastructure for communicating © information in machine-readable form. https://openfuture.eu/blog/a-first-look-at-the-copyright-relevant-parts-in-the-final-ai-act-compromise/
In this (relatively) recent piece, Venkat Rao compares generative AI systems to the Webb telescope, and argues that AI are not machines that produce something, but rather discover things. And the thing that they discover is information / intelligence that is inherent to data.
The argument - as often with Venkat's writing - often gets quite complicated. But the core argument is worth noting also for much less philosophical discussions about generative AI: that ulimately it's the data, and not the model that are crucial.
In the last months, I've been spending much time thinking about dataset governance and developing a commons-based framework for such governance. So Venkat's piece offers a useful theoretical underpinning, a story explaining why this is important.
There's been a lot of progress in 2023 on AI models, with dev teams playing the game of "who can count more billions of parameters?". It was also a year where there few positive developments in terms of dataset development and governance.
Hopefully, in 2024 this trend will reverse.
Venkat's piece:
https://studio.ribbonfarm.com/p/a-camera-not-an-engine
