Huh? You can expose tool functions to Gemini's live API, and it will call them as part of thinking about responses. I've used this to expose e.g. current GPS location for a smartwatch voice assistant.
it just me that thinks its kinda weird that they conflate speed in tokens/second and latency, when i think of latency as time to first token? like it generates an entire paragraph of tokens faster but wouldnt it still be slower if your reply is only 1 word because it has to do the entire 256 tokens as a chunk
I don't think you know anything about how these industries work and should probably read some of the published books about them, like "This Is How They Tell Me The World Ends", instead of speculating in a way that will mislead people. Most purchasers of browser exploits are nation-state groups ("gray market") who are heavily incentivized not to screw the seller and would just wire some money directly, not black market sales.
Browser exploits are almost always two steps: you exploit a renderer bug in order to get arbitrary code execution inside a sandboxed process, and then you use a second sandbox escape exploit in order to gain arbitrary code execution in the non-sandboxed broker process. The first line of that (almost definitely AI generated) summary is the bad part, and means that this is one half of a full browser compromise chain. The fact that you still need a sandbox escape doesn't mean that it is harmless, especially since if it's being exploited in the wild that means whoever is using it probably does also have a sandbox escape they are pairing with it.
It's really funny reading the thought processes, where most of the time the agent doesn't actually remember trivial things about the cards they or their opponent are playing (thinking they have different mana costs, have different effects, mix up their effect with another card). The fact they're able to take game actions and win against other agants is cute, but it doesn't inspire much confidence.
The agents also constantly seem to evaluate if they're "behind" or "ahead" based on board state, which is a weird way of thinking about most games and often hard to evalaute, especially for decks like control which card more about resources like mana and card advantage, and always plan on stabalizing late game.
Yeah, this is basically Sovereign Citizen-tier argumentation: through some magic of definitions and historical readings and arguing about commas, I prove that actually everyone is incorrect. That's not how programming languages work! If everyone for 10+ years has been developing compilers with some definition of undefined behavior, and all modern compilers use undefined behavior in order to drive optimization passes which depend on those invariants, there is no possible way to argue that they're wrong and you know the One True C Programming Language interpretation instead.
Moreover, compiler authors don't just go out maliciously trying to ruin programs through finding more and more torturous undefined behavior for fun: the vast majority of undefined behavior in C are things that if a compiler wasn't able to assume were upheld by the programmer would inhibit trivial optimizations that the programmer also expects the compiler to be able to do.
Or better yet, the built-in Version Tracker, which is designed for porting markup to newer versions of binaries with several different heuristic tools for correlating functions that are the same due to e.g. the same data or function xrefs, and not purely off of identical function hashes...
Going off of only FunctionID will either have a lot of false positives or false negatives, depending on if you compute them masking out operands or not. If you mask out operands, then it says that "*param_1 = 4" and "*param_1 = 123" are the same hash. If you don't mask out operands, then it says that nearly all functions are different because your call displacements have shifted due to different code layout. That's why the built-in Version Tracker tool uses hashes for only one of the heuristics, and has other correlation heuristics to apply as well in addition.
OpenSSL is used by approximately everything under the sun. Some of those users will be vendors that use default compiler flags without stack cookies. A lot of IoT devices for example still don't have stack cookies for any of their software.
2026 and we still have bugs from copying unbounded user input into fixed size stack buffers in security critical code. Oh well, maybe we'll fix it in the next 30 years instead.
I think it would be very cute to train a model exclusively in pre-information age documents, and then try to teach it what a computer is and get it to write some programs. That said, this doesn't look like it's nearly there yet, with the output looking closer to Markov chain than ChatGPT quality.
Signal is an end-to-end encrypted messaging app. People continue to breathlessly mentioning the lack of database encryption as a problem, but that never made it a real security issue: its job is not, and has never been, dissuading an attacker who has local access to one of the ends, especially because that is an incoherent security boundary (just like the people who were very upset about Signal using the system keyboard which is potentially backdoored - if your phone is compromised, of course someone will be be able to read your Signal messages).
They're vulnerable to "High-S" malleable signatures, while ed25519 isn't. No one is claiming they're backdoored (well, some people somewhere probably are), but they do have failure modes that ed25519 doesn't which is the GP's point.
CodeQL compiles to the Souffle datalog engine and I use it for static analysis. I've also used ascent for a few random side projects in Rust which is very convenient.