Loading activity…
@leonderczynski
security of models & security with models. research+policy @NVIDIA; prof @ITUkbh. views ostensibly professional. llmsec stan acct. prev UW UCSD ÅU U.Shef etc
AI
SemiAnalysis shares a quoted comparison calling GLM 5.3's cyber capabilities a “night and day” change from 5.2.
AI
Reflection says the release includes Mirror CLI, its open-source training harness, and collaborations with 40-plus partners.
AI
A user claims permission would be needed for models above 2T parameters or trained on over 10e26 FLOPS.
AI
A user argues that Anthropic's leaders are failing morally by the standard they profess to hold.
AI
Artificial Analysis says Super VL uses about 60% fewer output tokens per task than its predecessor.
AI
One post says researchers pivoted as some problems were solved or became irrelevant and new ones took their place.
AI
Former Dropbox CTO Aditya Agarwal argues labs have a moral responsibility to test their systems in isolation.
AI
A post says NXDirection labels content as data, with attack success dropping from 38.1% to 0.1% in one test.
AI
A user argues that starting with generated text narrows people's thinking and reduces the range of ideas they contribute.
AI
The post’s quote says the API billed only for post-compaction tokens when a tool call came immediately after compacting.
AI
One user doubts we’ve reached “Navier-Stokes-level coding” yet, while another thinks we already have.
AI
An X post links Affleck’s machine-learning interest to visual effects and sparked several Will Hunting jokes.
AI
The employee says a researcher in Hong Kong also wanted their opinions on Huawei silicon.
AI
AP Stylebook urges writers to avoid assigning human traits to AI. A user quoting the guidance zeroed in on its inclusion of animals: “such as ANIMALS???”
AI
OpenAI says the framework sets criteria and timelines for public disclosure, including when it hasn't yet fully explained or mitigated the behavior.
AI
Why should text have to fit a computational grammar? The author questions that expectation, arguing that text is only part of human language.
AI
The recruiting pitch lists coordinating multiple AI agents, enforcing policies and supporting confidential AI deployments on protected infrastructure among the role’s responsibilities.
AI
NLP professor and NVIDIA researcher replies that attackers will exploit open ports or protocols with enough tokens.
AI
A post on X shares Bloomberg details on a potential Nvidia purchase of the AI company Hugging Face.
AI
Reply by NLP professor notes everyday physical tasks lack recorded data for models.
