Mark Vero retweeted
Happy to have signed this excellent initiative!
"Europe cannot sleep through the AI revolution" - an open letter to 27 heads of state and government of EU, the President of the EC, and the President of the EC - the the54.eu/. Signed by a coalition of 59 leaders from science, technology, and business across member states of the European Union:
Thomas A. Henzinger
@HochreiterSepp
Georg Kopetz
@JoeyPrink
@stichris
@DeMoorBart
@hristoborisev
@mvechev
Marios Polycarpou
@StylianosLa
Silvio Kutic
@pechoucek
@JosefSivic
@SergeBelongie
@jepperindom
@jaakvilo
@samikaski
@petersarlin
Philippe Aghion
@olesovhcom
@ylecun
@FrankRHutter
Jarosław Kutyłowski
@bschoelkopf
Yannis Ioannidis
@moraitakis
@fhuszar
Dominik Mate Kovacs
Balázs Vinnai
@ricucch
@marcotrombetti
Noel E. O'Connor
@ScanlonPatricia
@AAmbainis
Artūrs Vasiļevskis
Agnė Paulauskaitė
Vytautas Savickas
@XavierBuck
@marktluszcz
@AngeloDalli
@alexieidingli
@CorinneVigreux
@wellingmax
@RBrzoska
@PiotrSankowski
@mati
Pedro Bizarro
@mariotelfig
@arlindooliveira
Marius Leordeanu
@mariabielik
@michalstencl
Sašo Džeroski
Nada Lavrač
Mark Pleško
@nuriaoliver
@OrusRoman
@antonosika
@DanicaKragic
Mark Vero retweeted
While people are preloading gta 6, Google dropped Gemini 4 Neon and achieved SOTA on our secure coding benchmark! Congrats to GDM!!
Results below, as well as cited on their launch post blog.google/innovation-and-a…
And fast-enough, the frontier in secure code generation is pushed further with Gemini 4 Neon. While Claude still tops the leaderboard on the easy and medium splits, Neon sets a new sota on hard, further opening the gap!
Congratulations to @GoogleDeepMind for the release!
And fast-enough, the frontier in secure code generation is pushed further with Gemini 4 Neon. While Claude still tops the leaderboard on the easy and medium splits, Neon sets a new sota on hard, further opening the gap!
Congratulations to @GoogleDeepMind for the release!
Leaderboard: baxbench.com/autobaxbuilder/…
33
🏅How secure is code generated by the latest frontier models? 🏅
After a long hiatus, we decided to see where the current frontier lies in secure code generation. While the models have improved massively, to much of our surprise, they still generate a lot of vulnerabilities.
While it is great to see models improving at secure code generation, there remains a lot of responsibility on the developer and code owners to ensure the security of the final code. With the increasing ease of exploiting vulnerabilities, these improvements are critical!
1
1
96
See the full leaderboard here: baxbench.com/autobaxbuilder/…
Relied on great work by everyone @tobiasvonarx_, @nielstron, @baader_max, @Chenhao3564, and @mvechev at the @the_sri_lab!
1
107
Turns out if your code base is full of security vuls, agents will just introduce more issues
🦠 Coding agents propagate vulnerabilities in code bases!
How comes?
When a repo already contains vulnerable code, they write similar vulnerabilities into new code, even when the vulnerable code is functionally unrelated to their task and they can detect its flaw.
2
68
Mark Vero retweeted
🦠 Coding agents propagate vulnerabilities in code bases!
How comes?
When a repo already contains vulnerable code, they write similar vulnerabilities into new code, even when the vulnerable code is functionally unrelated to their task and they can detect its flaw.
Mark Vero retweeted
Does more reasoning improve performance on LeanLean (disproportionally)?
It turns out that the model capability trend is still log-scaling.
We have now evaluated Sol 6.1 and Opus 5.5 across multiple reasoning efforts on leanleanbench.com!
Mark Vero retweeted
my colleagues found out our coffee machine is controllable via Bluetooth and now assessing which agent brews the best coffee
Mark Vero retweeted
From day one, mimic has been focused on a single goal: general-purpose dexterous manipulation. Today we're proud to announce the mimic hand M1 and the mimic wearable U1.
We believe the only way to solve dexterous manipulation at scale is by going full-stack at the frontier of physical AI, building every layer ourselves around one fixed point, the human hand.
The M1 is a highly backdrivable, tendon-driven hand that covers the full range of human capability, from heavy payloads to fine manipulation.
If you are at ICML check out our oral presentation on Saturday at 11AM the DEMO workshop about adapting reasoning models with cheap and simple techniques.
(DEMO @ #ICML26 Oral)
RLMs are undoubtly powerful. But how can we train them further?
IFT on golden answers can destroy reasoning capabilities.
We show: With our careful merging technique, reasoning can be restored entirely, while keeping gains on the target domain! 🧵
2
143
Mark Vero retweeted
(DEMO @ #ICML26 Oral)
RLMs are undoubtly powerful. But how can we train them further?
IFT on golden answers can destroy reasoning capabilities.
We show: With our careful merging technique, reasoning can be restored entirely, while keeping gains on the target domain! 🧵




















