Vincent Touquet

25.2K posts

Vincent Touquet

Vincent Touquet

@vinstar

Katılım Mayıs 2007
4.4K Takip Edilen1.1K Takipçiler
Vincent Touquet retweetledi
Evis Drenova
Evis Drenova@evisdrenova·
Claude Shannon outlined what problems are “good ones” for AI to solve 70 years ago.
Evis Drenova tweet media
English
4
1
5
252
Sam Coates Sky
Sam Coates Sky@SamCoatesSky·
This is a brilliant - bleak - piece about the broken local politics of Birmingham. First class on the ground reporting by @alexrogerssky @JoshGafson1 These pieces aren’t easy - you only get footage like this with skill, graft and luck - please watch:
English
197
337
1.5K
298.8K
Vincent Touquet
Vincent Touquet@vinstar·
@thekitze how long until clawd / claude /... in the dictionary like google ? 🤔
English
0
0
0
121
kitze 🛠️ tinkerer.club
claude turned into a cultish religion thingie that openai will NEVER achieve no matter what they ship. it's a vibe and a lifestyle. (saying this as someone who recently cancelled their sub and is using codex since last year)
English
30
1
108
7.1K
Vincent Touquet retweetledi
Raffi Hotter
Raffi Hotter@raffi_hotter·
This algorithm uses one of my favourite theorems in math, the Johnson-Lindentrauss Lemma, which says you can drastically reduce the dimensionality of n points to just log(n) dimensions and still preserve pairwise distances
Google Research@GoogleResearch

Introducing TurboQuant: Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency. Read the blog to learn how it achieves these results: goo.gle/4bsq2qI

English
32
216
3.1K
232.6K
Mike Mihajlovic
Mike Mihajlovic@MihajlovicMike·
Recent intelligence reports indicate that the airspace of Poland, Lithuania, Latvia, and Estonia is being utilized to facilitate Ukrainian drone strikes targeting oil and gas infrastructure in the Leningrad Oblast. Additionally, there are indications that operational coordination for these attacks is occurring from Finnish territory. The scale and scope of these operations suggest a deep level of involvement by NATO countries, with little attempt to conceal participation. From a strategic perspective, this situation represents a significant escalation in the conflict, as it involves the active use of third-party airspace to strike critical Russian energy infrastructure. The involvement of multiple NATO states in facilitating or supporting these operations, even indirectly, raises the potential for broader geopolitical consequences. It also underscores the increasing role of unmanned aerial systems in modern warfare, particularly as tools for precision strikes that can bypass traditional frontlines. The implications for regional security are substantial. The use of NATO-controlled airspace for attacks on Russian territory may exacerbate tensions and increase the risk of miscalculation or unintended confrontation. At the same time, the attacks on energy infrastructure have the potential to create economic and logistical disruptions beyond the immediate theater of operations, affecting energy supply chains and regional markets. In this context, the conflict is no longer limited to conventional troop movements or localized engagements; it now includes complex, multi-domain operations involving coordination across international boundaries. The apparent lack of concealment regarding NATO’s role in supporting these drone operations highlights a shift toward a more overt form of strategic competition, with long-term implications for European security architecture and military doctrine.
English
16
76
208
8K
Vincent Touquet retweetledi
Ben
Ben@jt_martin·
the attacker compromised a security scanning tool first, then used those credentials to poison the AI package. and the only reason it got caught is because the malware was written so badly it crashed machines. if the attacker had been 10% more competent, this runs for months and we're reading about it in an SEC filing instead of a tweet
English
2
11
129
26.2K
Vincent Touquet retweetledi
Aakash Gupta
Aakash Gupta@aakashgupta·
Someone just poisoned the Python package that manages AI API keys for NASA, Netflix, Stripe, and NVIDIA.. 97 million downloads a month.. and a simple pip install was enough to steal everything on your machine. The attacker picked the one package whose entire job is holding every AI credential in the organization in one place. OpenAI keys, Anthropic keys, Google keys, Amazon keys… all routed through one proxy. All compromised at once. The poisoned version was published straight to PyPI.. no code on GitHub.. no release tag.. no review. Just a file that Python runs automatically on startup. You didn’t need to import it. You didn’t need to call it. The malware fired the second the package existed on your machine. The attacker vibe coded it… the malware was so sloppy it crashed computers.. used so much RAM a developer noticed their machine dying and investigated. They found LiteLLM had been pulled in through a Cursor MCP plugin they didn’t even know they had. That crash is the only reason thousands of companies aren’t fully exfiltrated right now. If the code had been cleaner nobody notices for weeks. Maybe months. The attack chain is the part that gets worse every sentence. TeamPCP compromised Trivy first. A security scanning tool. On March 19. LiteLLM used Trivy in its own CI pipeline… so the credentials stolen from the SECURITY product were used to hijack the AI product that holds all your other credentials. Then they hit GitHub Actions. Then Docker Hub. Then npm. Then Open VSX. Five package ecosystems in two weeks. Each breach giving them the credentials to unlock the next one. The payload was three stages.. harvest every SSH key, cloud token, Kubernetes secret, crypto wallet, and .env file on the machine.. deploy privileged containers across every node in the cluster.. install a persistent backdoor waiting for new instructions. TeamPCP posted on Telegram after: “Many of your favourite security tools and open-source projects will be targeted in the months to come.. stay tuned.” Every AI agent, copilot, and internal tool your company shipped this year runs on hundreds of packages exactly like this one… nobody chose to install LiteLLM on that developer’s machine. It came in as a dependency of a dependency of a plugin. One compromised maintainer account turned the entire trust chain into a credential harvesting operation across thousands of production environments in hours. The companies deploying AI the fastest right now have the least visibility into what’s underneath it.
Andrej 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.

English
282
2.2K
10.8K
2.6M
Vincent Touquet retweetledi
Rudy Havenstein, Senior Markets Commentator.
Sound advice from Tony Deden: "Governments and politicians lie. The press lies. Treating official narratives as solid ground, especially in war, is a mistake. In wartime, truth is obscured by design."
Rudy Havenstein, Senior Markets Commentator. tweet media
English
5
38
108
8.6K
Vincent Touquet retweetledi
shirish
shirish@shiri_shh·
THE APPLE APP STORE IS DROWNING IN AI SLOP people are treating the App Store like a Medium blog spitting out apps one after another. All with zero users and $0 revenue. Apple reviews that used to take hours are now stretching into WEEKS and even months > more than 550k apps were submitted just last year, highest in a decade.
shirish tweet media
English
378
345
3.2K
529.9K
Vincent Touquet retweetledi
Anna Riedl
Anna Riedl@AnnaLeptikon·
The real question of our time is not whether an artificial intelligence explosion is underway, but how it interacts with a simultaneous natural intelligence implosion. I'd argue everything technical remains in the end causally grounded in a stable substrate of human intelligence, both individual and collective as a civilization. If that soil deteriorates, so will the technological stack that grows from it, which is ontologically to be understood as being in a higher organizational thicket and causally dependent on the integrity of lower organizational entities. And yes, there is no way past the question of energy once we speak of organizational thickets: Complexification, the existence of higher organizational entities with both more diversification and integration, inherently requires more energy in the system. A majority of people "forget" this simple truth (deterioration of aggregated individual intelligence), and those who stay grounded in reality are no longer in charge (deterioration of collective intelligence). This energy blindness through intelligence implosion is tearing down civilization even more thoroughly. When I say "forgetting" you have to keep in mind that forgetting is the default: every new generation is born irgnorant and if they are not taught important lessons, they don't know them. The information is then collectively forgotten. The maintenance of knowledge and information in society is as crucial as the maintenance of physical infrastructure. It is to be understood as our most crucial infrastructure, and the ability to maintain anything is one of the most important lessons in need of maintenance.
Anna Riedl tweet mediaAnna Riedl tweet mediaAnna Riedl tweet media
Selective Breeding and the Birth of Anime@SelectivBreeder

"you will live long enough to see the intelligence implosion in the global North decimate 90% of the global South"

English
6
10
49
3.7K
Open Source Intel
Open Source Intel@Osint613·
Hezbollah released a video claiming it targeted the Shayetet 13 naval base at Atlit, south of Haifa. The video, however, shows nothing to back the claim. Just launches and B-roll of rockets.
English
5
26
172
35.1K
Vincent Touquet retweetledi
Martin Varsavsky
Martin Varsavsky@martinvars·
🚨 AI is moving faster than regulators can comprehend. In medical AI, we are seeing breakthroughs that will completely transform healthcare in the next 36 months. Europe needs to deregulate and embrace this, or sit on the sidelines while the US and Israel build the future.
English
2
3
29
2.1K
Vincent Touquet retweetledi
Dwarkesh Patel
Dwarkesh Patel@dwarkesh_sp·
When Copernicus proposed heliocentrism in 1543, it was actually less accurate than Ptolemy's geocentric model - a system refined over 1,400 years with epicycles precisely tuned to match observed planetary positions. It took another 70 years before Kepler, working from Tycho Brahe's unprecedentedly precise observations, replaced Copernicus’s circles with ellipses - finally making heliocentrism empirically superior. Terence Tao's point is that science needs a high temperature setting. If we only fund and follow what's most state of the art today, we kill the ideas that might need decades of work to surpass some overall plateau.
English
122
587
4.8K
527.9K