Guillaume Ceccarelli (@GCsVentures) profile picture

Rebooting Human. Agency maximalist, BS & bureaucracy antagonist. Respects adversity, growth, erudition, generosity. Not status, position or ideology. #RWRI alum

Joined August 2010
Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)Photo by Guillaume Ceccarelli (@GCsVentures)
Very few things I love more than human potential, and very few things I find more painful than the destruction of it.
1
2
16
3,311
GPT-6: Ad astra per obscura
you have to be kidding me
240
Guillaume Ceccarelli retweeted
Replying to @PessimistsArc
Right. Dario was already claiming that GPT2 was too dangerous to open source back in 2019. I made fun of them then. Everyone should make fun of them now.
390
3,163
464
25,870
1,996,460
I mean... yeah. Also great anti-ad for Azure right there.
This is so funny
Readers added context they thought people might want to know
GitHub launched a new status page in December 2018 that introduced granular reporting of individual component uptimes instead of overall service uptime, making pre- and post-2018 data not directly comparable. github.blog/engineering/in… statusgator.com/blog/has-githu…
152
On behalf of Europeans everywhere, I would like to thank the great state of New Jersey for making us feel less alone, by adopting legal requirements on par with our EU elite's level of intellectually challenged. It's very moving. Thank you.
NJ, king of stupid laws
2
127
Guillaume Ceccarelli retweeted
WATERMARK STRIPPER HITS 10,000 STARS DAYS AFTER ANTHROPIC SHIPS PROVENANCE Anthropic detailed how Claude's text watermark works last week. By this morning a MIT-licensed repo that strips it has 10,000 stars, targeting Claude, SynthID-Text and OpenAI marks, plus C2PA and EXIF in images and PDFs. github.com/guillaumemeyer/wa…
Claude to embed watermarks in ALL OUTPUT TEXT
58
123
54
1,032
596,044
Do you have a license for that printing press? /s And since text watermarking functions by introducing a statistical bias in token selection, EU essentially mandated model extra stupidity / product enshittification. Another EU reg good for bureaucrat self-congratulation only.
“I don’t live in EU, why should I have my AI output watermarked?” because EU law basically says: you can generate a text in your cozy SF cafe, but if it gets used in EU, some dude there must be able to detect if it was AI. welcome to the regulation party, everyone is invited
1
132
Other than EU bureaucrats for their self-enjoyment, who else benefits? Non-compliant model makers and non-compliant model users At best, it makes the AI ACT eventually unenforceable, and at worst that makes the EU poorer through furthering economic irrelevance and isolation 🇪🇺👏
1
22
Some people believe that the EU is generally good, and it only needs competent people at the helm. The problem is that the system is itself built to keep people isolated from decision-making power, thus keeping power disconnected from ground realities. To fix it is to break it.
9
With all of Anthropic's talk about regulating AI, one regulation I could get behind would be for them to be required to make their scanned PDF collections available. Especially the 1/1 book editions. It should be a crime to destroy rare texts on an industrial scale to begin with.
btw anthropic's internal document on this literally said "we don't want it to be known that we are working on this.” it was called project panama. here's exactly what happened: 1: anthropic concluded that books were the cheapest way to build a world-class model because they gave claude curated facts, structured arguments, compelling stories, and writing “an editor would approve of.” 2: once anthropic decided it needed books at enormous scale, its first solution was piracy. it downloaded 7m+ books from online libraries including libgen. the judge later wrote that although anthropic had legal ways to buy them, it chose piracy to avoid what dario amodei called the “legal/practice/business slog.” 3: that piracy created a massive legal risk. so in february 2024, anthropic hired tom turvey, the former head of partnerships for google books, to find a legally safer way of obtaining “all the books in the world.” 4: turvey first contacted major publishers about licensing their catalogs. those attempts didn’t produce agreements, so anthropic chose a route that required no publisher permission: buying millions of physical books through distributors and used-book retailers. 5: within about a year, anthropic spent tens of millions acquiring and scanning millions of books, including many rare and 1/1 titles. one vendor proposal targeted 500,000 to 2 million books in six months. 6: to scan that many books within months, the vendors physically dismantled them. a hydraulic cutter removed each spine. the pages were trimmed to size, fed as loose sheets through high-speed industrial scanners, and converted into searchable PDFs. the paper remains were then sent for recycling. 7: these PDFs were fed into claude as training data. the complete collection became a private, searchable anthropic library that the company planned to “store forever.” the scans aren’t available to the public and were never open-sourced.
1
1
9
843
As the communication around Mythos & Fable has been variations of "OMG so powerful and dangerous!", mixed with "safety" and "control", this reads as consequences + reinforcement of the same. Best thing for Anthropic (and users!) at this point is to get acquired or out-competed.
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…
1
113
Signal is audited, Open Source and encryption keys for your messages reside on your devices & nowhere else. X Chat isn’t and encryption keys for your messages reside on their hardware. So… use Signal. + If they mess up, it will make a lot of noise very fast. X chat? Less likely.
This quoted post is unavailable.
4
322
You could think that nothing can steal your sensibility, your unique way to see, take in, and create from the world. What makes you human at your core. Your inner poetry, if you will. And you’d be wrong. As all things, it can break. Waste no time + care for your mind & heart.
1
151
This is why we should not debase the meaning of ”open source”. One team will say, ”don’t compete with us”, another ”don’t make too much money”, another still ”only if you run a few machines”. Next thing you know, ”open source” means nothing anymore. License restricts use? Not OS
"Modified MIT" license: "You are not authorized to exercise any rights under this license if the global consolidated monthly revenue of your company (or that of your employer) exceeds $20 million…"
1
2
3,752
Why does a medical journal post about politics? Unprofessional & this same attitude behind (mis)handling the pandemic caused the cratering trust in medical institutions. Some MDs here also constantly post about politics. Do you not see how you're undermining your legitimacy?
Far right parties are gaining ground across the world. The more that governments seem unable to improve the material conditions of people’s lives, the greater the risk that people will look elsewhere for solutions, says @COPDdoc bmj.com/content/391/bmj.r251…
2
1
1
9
880
6 years of gaslighting & incompetence in the highest spheres, with an absence of indictments of the people responsible for the research which led to the pandemic, the lives lost & altered. There's never been clearer evidence of a global lack of accountability where it matters.
6 years ago today: A man in Wuhan, China starts feeling ill, becoming the first confirmed case of COVID-19
1
90
Universal Income AI takes fail to take economics & human nature into account. We are hardwired for purpose. Purpose involves causing changes and leveraging resources to better your life and that of those you care about. Resources imply wealth and transfers. Changes imply work.
There will be no rich people in the future. Everyone will receive UEI payments worth many times the value of today’s average salary, while AI and robots provide goods and services at the highest possible quality. You won’t be able to buy better food, health care, or transportation with more money. Owning a bigger house will not make life better. Excess will be looked down upon. Materialism will be shunned - seen as shallow, selfish, and out of touch. People will admire simplicity and humanity. Real wealth will be found in having a loving family, strong social relationships, and a solid moral foundation. The biggest transformation will not be in the economy - it will be in ourselves.
1
1
1
221
And if what you do is found valuable by others, congratulations: some might be willing to benefit in exchange for… resources: you just reinvented entrepreneurship, income and capitalism. AI does not make atoms free or transforming them (i.e. energy) free or experiences scalable.
71
Never confuse a belief one adopts so as to execute to the best of their abilities, with an enlightened, let alone accurate opinion about what the future will actually look like.
🚨Anthropic researcher believes: “maybe as soon as the first half of next year: software engineering is done.” ITS SO OVER.
1
238
What surprises me most with the e-invoicing / PEPPOL mandates, is how most seem cool with it. As if adding extra chains of dependencies, rent-extracting gatekeepers and technical points of failure for basic business operations, fragilizing everything in the process, is just fine
120
Respect and reward competence, get competence more. Respect and reward other things, get other things more. Glorifying assholes was never the point, but competent people, staking themselves and cutting through the BS have always been valuable leaders. Re-normalize competence.
What the HECK is going on with tech? In the last week: Multiple cloud outages, x DMs totally broken, antigravity doesn't work, my watch is showing me 15 year old cal events, mac OS is a mess, email is spammed to hell and every nerd on here is talking like new AI is the second coming
1
1
2
157
In fact, I think part of why new AI models are being treated like a second coming is precisely because we are craving for access to competence. It’s not just that technology is offering further know-how, it’s also that irrelevant gatekeepers are further less able to slow us down.
2
68