Submit an article
thehtml.review
the html review 05
by mishaderidder.eth11868 🥝16h
@karpathy
@karpathy

Software horror: litellm PyPI supply chain attack. Simple `pip install litellm` was enough to exfiltrate SSH keys, AWS/GCP/Azure creds, Kubernetes configs, git credentials, env vars (all your API keys), shell history, crypto wallets, SSL private keys, CI/CD secrets, database passwords. LiteLLM itself has 97 million downloads per month which is already terrible, but much worse, the contagion spreads to any project that depends on litellm. For example, if you did `pip install dspy` (which depended on litellm>=1.64.0), you'd also be pwnd. Same for any other large project that depended on litellm. Afaict the poisoned version was up for only less than ~1 hour. The attack had a bug which led to its discovery - Callum McMahon was using an MCP plugin inside Cursor that pulled in litellm as a transitive dependency. When litellm 1.82.8 installed, their machine ran out of RAM and crashed. So if the attacker didn't vibe code this attack it could have been undetected for many days or weeks. Supply chain attacks like this are basically the scariest thing imaginable in modern software. Every time you install any depedency you could be pulling in a poisoned package anywhere deep inside its entire depedency tree. This is especially risky with large projects that might have lots and lots of dependencies. The credentials that do get stolen in each attack can then be used to take over more accounts and compromise more packages. Classical software engineering would have you believe that dependencies are good (we're building pyramids from bricks), but imo this has to be re-evaluated, and it's why I've been so growingly averse to them, preferring to use LLMs to "yoink" functionality when it's simple enough and possible.

x.com
here.now
here.now | Instant web hosting for agents
by timdaub.eth11790 🥝8h
openwallet.sh
OWS - Open Wallet Standard
by timdaub.eth11790 🥝2d
@leventnacakci
@leventnacakci

Refik Anadol’s work frequently employs terms like “fluid dynamics,” “latent space,” and “algorithmic brushstrokes,” which give the viewer a false sense of intellectual depth. But the reality is this: these terms often serve to mask the artist’s aesthetic choices behind a façade of scientific authority. When the artist says, “I am visualizing bird sounds recorded in a forest,” the viewer is led to believe that the resulting image is a natural and objective representation of those sounds. However, as discussed earlier, the direction of the flow or the choice of color is entirely arbitrary. In a scientific experiment, input A should consistently produce output B, or at least the relationship should be demonstrable. Here, however, input A (bird sound) can be transformed into color C or motion D depending on the artist’s mood that morning. This is not an experiment; it is simply decoration. Phrases like “artificial intelligence is dreaming” or “data sculpture” mystify technical processes. Calling a pixel-generation process based on statistical probabilities “dreaming” is nothing more than presenting a basic mathematical regression as something magical. This causes the work to derive its power not from its own aesthetics, but from the “coolness” of science and technology. If we were to replace the bird sounds with the sound of a vacuum cleaner using the same “mapping” settings, we would still obtain a “mesmerizing” visual. In other words, the beauty of the image does not come from the essence of the data, but entirely from the artist’s graphics engine. In this context, extracting data from bird sounds is not a technical necessity, but rather a storytelling device—a marketing element of the project. Wrapping data in a scientific veneer reinforces the illusion that the work is “meaningful.” But once this illusion dissolves, what remains is merely a high-resolution “screensaver.” Real science uses data to understand reality; this kind of “art,” by contrast, uses data merely as spectacle. Instead of saying, “I shaped this data according to my own will and created something beautiful,” the artist claims: “Through scientific algorithms, I revealed the hidden architecture of the data.” This rhetoric is nothing more than putting on a mask of scientific authority to influence the viewer. This form of pseudo-scientific framing is monetized; data is dramatized, the viewer is drawn into a sense of technological awe, and reality is obscured through spectacle.

x.com
imagedelivery.net
humans aren‘t the users anymore
by timdaub.eth11790 🥝1d