Some #generativeAI developers love to destroy the foundations of the tech they build. #WIkipedia is one of the most valuable sources of genAI training data. Undermining it is not just attacking a great common resource. It's also completely self-destructive arstechnica.com/ai/2026/01/n...
A little-known nonprofit has been lying to news publishers while funneling millions of paywalled articles to tech companies for AI training. Read my investigation in The Atlantic. www.theatlantic.com/technology/2...
Check in if you're interested in my thoughts about what open source AI should aspire to be in relation to proprietary AI
"The update is yet another signal that payment processors...are currently the ultimate arbiter of what kind of content can be made easily available online, or not."
The key questions we always should ask when people talk about AI: What is being automated and why? @alexhanna.bsky.social @weizenbauminstitut.bsky.social
"AI is a labor disciplining device" @alexhanna.bsky.social
“The reporter is a man of critical value. No amount of money or effort spent in fitting the right men for this work could possibly be wasted, for the health of society depends upon the quality of the information it receives.” — Walter Lippmann [a century later, I’d swap “man” for “person” though]
New Release! Most AI deepfakes aren't political. 90% of deepfakes are non-consensual intimate imagery. 99% of victims are women. Max Hoppensted, @rechercheur.bsky.social, @romanhoefner.bsky.social, and I uncover a deepfake community and the business behind undress apps www.spiegel.de/netzwelt/web...
"brainstorming and iteration is...a crucial everyday part of game development...and is not a problem to be solved...I have had many discussions with other game developers who interact with AI engineers and savants who believe our industry pipelines need 'fixing' by them and them alone"
Die Union will das Informationsfreiheitsgesetz abschaffen.
@arnesemsrott.bsky.social: „Öffentliche Kontrolle &Transparenz sind der Union offenbar ein Dorn im Auge. Sie will unbehelligt durchregieren. Rechte der Öffentlichkeit stören dabei offenbar."
Pressemitteilung: fragdenstaat.de/newsletter/a...
«By moving fast and breaking things, DOGE forces a collapse of the system where unanswered questions are met with technological solutions. Shifting the conversation to the technical is a way of locking policymakers and the public out of decisions and shifting that power to the code they write.»
You can’t post your way out of fascism
Authoritarians and tech CEOs now share the same goal: to keep us locked in an eternal doomscroll instead of organizing against them
🔗 www.404media.co/you-cant-pos...
A bird's-eye view of a former Auschwitz II-Birkenau camp showing a wide dirt pathway flanked by parallel rows of barbed-wire fences. Groups of visitors walk along the path, surrounded by the remnants of brick structures and barracks, now reduced to foundations. Green grass contrasts with the somber history of the site, as the path leads toward a guard tower in the distance.
Auschwitz was at the end of a long process. It did not start from gas chambers.
This hatred was gradually developed by humans. From ideas, words, stereotypes & prejudice through legal exclusion, dehumanization & escalating violence... to systematic and industrial murder.
Auschwitz took time.
“AI is fake and sucks” vs “AI is real and dangerous” is a Twitter argument. In reality I think the debate also has a lot of “AI is real but not for how you’re using it,” to “AI is fake and that is dangerous,” to “things are happening to real people because of AI hype and that should stop.”
My reading for this week, delivered to me by the great
@aschrock.bsky.social
themself! Thank you, looking forward to reading :-)
Labelers training AI say they're overworked, underpaid and exploited by big American tech companies
Dieser Report gibt Hoffnung!
Immer mehr neue, ambitionierte Medien haben sich in Deutschland und Europa gegründet. Medien mit dem Ziel, die Öffentlichkeit hochwertig zu informieren.
@netzwerkrecherche.org hat für den „Journalism Value Report“ 174 Medien in 31 Ländern befragt und kann zeigen:
I have a new piece out with @aisvarya17.bsky.social in @columjournreview.bsky.social in which we test how OpenAI's new search feature surfaces and attributes news content. Our findings were not promising for news publishers (1/9) www.cjr.org/tow_center/h...
“Without facts, you can’t have truth, and without truth, you can’t have trust”. - Maria Ressa, 2021 Nobel Peace Prize
The Onion should buy Elsevier next
It ended well though. He got the job, and still has it. We met recently 😅
I still remember when a friend asked for advice about getting a job I intended to apply for
Long term, there should be less reliance on sources like Common Crawl and a bigger emphasis on training generative AI on datasets created and curated by people in equitable and transparent ways (10/10)
A key issue is that filtered Common Crawl versions are not updated after their original publication to take feedback and criticism into account. Therefore, we need dedicated intermediaries tasked with filtering Common Crawl in transparent and accountable ways that are continuously updated (9/10)
AI builders should put more effort into filtering Common Crawl, establish industry standards and best practices for end-user products to reduce potential harms when using Common Crawl or similar sources for training data (8/10)
Both Common Crawl and AI builders can help making generative AI less harmful. Common Crawl should highlight the limitations and biases of its data, be more transparent and inclusive about its governance, and enforce more transparency by requiring AI builders to attribute using Common Crawl (7/10)
Due to Common Crawl’s deliberate lack of curation, AI builders need to filter it with care, but such care is often lacking. Popular filtered versions like C4 are especially problematic as the filtering techniques used to create them are simplistic and leave lots of harmful content untouched (6/10)
In addition, relevant domains like Facebook and the New York Times block Common Crawl from crawling most (or all) of their pages. These blocks are increasing, creating new biases in the crawled data www.wired.com/story/most-n... (5/10)
Common Crawl archive is massive, but far from being a “copy of the internet.” Its crawls are automated to prioritize pages on domains that are frequently linked to, making digitally marginalized communities less likely to be included. Moreover, most captured content is English (4/10)
Using Common Crawl's data does not easily align with trustworthy and responsible AI development because Common Crawl deliberately does not curate its data. It doesn't remove hate speech, for example, because it wants its data to be useful for researchers studying hate speech (3/10)
