Dems of this ilk, who have been in charge of the party my whole life, have been crystal clear that they see their relationship with the GOP as one of power sharing. They absolutely do not want to be in the position of adversarially taking the GOP down. That’s what this is, as it has ever been.
November 4, 2026 at 2:36 PM
Dems of this ilk, who have been in charge of the party my whole life, have been crystal clear that they see their relationship with the GOP as one of power sharing. They absolutely do not want to be in the position of adversarially taking the GOP down. That’s what this is, as it has ever been.
We need to get a 2028 candidate with transformative goals, if they're going to have the stature of FDR+Reagan (note for people adversarially reading, I am not saying Reagan was good, just popular in the 80s, so much so that it traumatized a generation of Democrats)
November 6, 2026 at 8:20 PM
We need to get a 2028 candidate with transformative goals, if they're going to have the stature of FDR+Reagan (note for people adversarially reading, I am not saying Reagan was good, just popular in the 80s, so much so that it traumatized a generation of Democrats)
I swear they adversarially refuse to answer softball questions in a way that is designed to make me want to quit my job
November 4, 2026 at 3:03 PM
I swear they adversarially refuse to answer softball questions in a way that is designed to make me want to quit my job
I dated this woman once and she was real educated right like from a real big famous school and she got a great big heavy degree from that place and she was always pulling this nonsense it's like do you just approach everything adversarially some of us just aren't wired like that you know
I have never understood why people think otherwise. It is literally just saying "oh hey yeah I know how that feels" but neurotypical people love getting insulted for no reason
November 5, 2026 at 5:22 PM
I dated this woman once and she was real educated right like from a real big famous school and she got a great big heavy degree from that place and she was always pulling this nonsense it's like do you just approach everything adversarially some of us just aren't wired like that you know
A variant of this I've found useful is welding red-green development to input fuzzing to adversarial review. Make the agent adversarially review tests as well as code, dropping or fixing any that don't map to a functional or security property, and randomizing wherever possible.
November 7, 2026 at 1:27 PM
A variant of this I've found useful is welding red-green development to input fuzzing to adversarial review. Make the agent adversarially review tests as well as code, dropping or fixing any that don't map to a functional or security property, and randomizing wherever possible.
i think small imprecision gets rounded away by the horse's cognitive processes, the horse's decision flowchart isn't changing round to round so even if the exact details fudge the numbers from p(it goes left)=0.7 to p(left)=0.6 you can still adversarially manipulate its behaviour
November 4, 2026 at 8:21 PM
i think small imprecision gets rounded away by the horse's cognitive processes, the horse's decision flowchart isn't changing round to round so even if the exact details fudge the numbers from p(it goes left)=0.7 to p(left)=0.6 you can still adversarially manipulate its behaviour
RESIST: Resilient Decentralized Learning Using Consensus Gradient Descent
Cheng Fang, Rishabh Dixit, Waheed U. Bajwa, Mert Gurbuzbalaban
Action editor: Han Bao
https://openreview.net/forum?id=aNu8S9O6Ug
#adversarially #adversarial #attack
Cheng Fang, Rishabh Dixit, Waheed U. Bajwa, Mert Gurbuzbalaban
Action editor: Han Bao
https://openreview.net/forum?id=aNu8S9O6Ug
#adversarially #adversarial #attack
November 5, 2026 at 5:21 PM
RESIST: Resilient Decentralized Learning Using Consensus Gradient Descent
Cheng Fang, Rishabh Dixit, Waheed U. Bajwa, Mert Gurbuzbalaban
Action editor: Han Bao
https://openreview.net/forum?id=aNu8S9O6Ug
#adversarially #adversarial #attack
Cheng Fang, Rishabh Dixit, Waheed U. Bajwa, Mert Gurbuzbalaban
Action editor: Han Bao
https://openreview.net/forum?id=aNu8S9O6Ug
#adversarially #adversarial #attack
Chengzhi Guo, Shaofeng H. C. Jiang, Robert Krauthgamer, David P. Woodruff, Shenghao Xie, Samson Zhou: Adversarially Robust Streaming Algorithms for $(k, z)$-Clustering https://arxiv.org/abs/2611.07678 https://arxiv.org/pdf/2611.07678 https://arxiv.org/html/2611.07678
November 6, 2026 at 6:40 AM
Chengzhi Guo, Shaofeng H. C. Jiang, Robert Krauthgamer, David P. Woodruff, Shenghao Xie, Samson Zhou: Adversarially Robust Streaming Algorithms for $(k, z)$-Clustering https://arxiv.org/abs/2611.07678 https://arxiv.org/pdf/2611.07678 https://arxiv.org/html/2611.07678
I agree with this and I think one of the biggest problems with it is legal risks of everything you say being ever being used adversarially against you in discovery
November 4, 2026 at 11:19 PM
I agree with this and I think one of the biggest problems with it is legal risks of everything you say being ever being used adversarially against you in discovery
im having opus create an ETL pipeline and basically just throwing tokens at it to "audit" it adversarially and make it more robust and im afraid that its working quite well. its so damn Weird.
I hate how much this is increasingly describing my life.
November 5, 2026 at 8:30 PM
im having opus create an ETL pipeline and basically just throwing tokens at it to "audit" it adversarially and make it more robust and im afraid that its working quite well. its so damn Weird.
Policy lesson: think adversarially. Platforms over-comply and grifters exploit the fallout. Takedowns need public, auditable traces to specific legal clauses, and neutral transparency rules should be decoupled from subjective “harms” ones.
#misinfo #onlineSafety #techPolicy #comms #regulation #UK
#misinfo #onlineSafety #techPolicy #comms #regulation #UK
The Only Way From Essex: Strategic Lessons in Narrative Warfare
When it comes to information integrity and the state, “it's like déjà vu all over again”
sohandsouza.substack.com
November 2, 2026 at 3:33 PM
Policy lesson: think adversarially. Platforms over-comply and grifters exploit the fallout. Takedowns need public, auditable traces to specific legal clauses, and neutral transparency rules should be decoupled from subjective “harms” ones.
#misinfo #onlineSafety #techPolicy #comms #regulation #UK
#misinfo #onlineSafety #techPolicy #comms #regulation #UK
But this is the thing: modern “agentic” LLM workflows explicitly encourage you to remove all human review from the loop, and rely on orchestrating a bunch of LLM agents to generate and adversarially review and fix up code.
November 2, 2026 at 1:59 AM
But this is the thing: modern “agentic” LLM workflows explicitly encourage you to remove all human review from the loop, and rely on orchestrating a bunch of LLM agents to generate and adversarially review and fix up code.
as i recall the initial Jepsen analyses for MongoDB primarily concerned their consistency issues under replication, right?
ParticleDB failing single-node like this is, as far as i can tell, markedly worse. it implies people failed to test it themselves at all, not just adversarially.
ParticleDB failing single-node like this is, as far as i can tell, markedly worse. it implies people failed to test it themselves at all, not just adversarially.
November 2, 2026 at 12:44 AM
as i recall the initial Jepsen analyses for MongoDB primarily concerned their consistency issues under replication, right?
ParticleDB failing single-node like this is, as far as i can tell, markedly worse. it implies people failed to test it themselves at all, not just adversarially.
ParticleDB failing single-node like this is, as far as i can tell, markedly worse. it implies people failed to test it themselves at all, not just adversarially.
The Scissors Effect: When Resize-Based Input Diversity Helps or Hurts Transfer Attacks
Yuhang Jiang, Xiaojing Chen
Action editor: Amartya Sanyal
https://openreview.net/forum?id=b4pCcgJM0M
#adversarially #attacks #attack
Yuhang Jiang, Xiaojing Chen
Action editor: Amartya Sanyal
https://openreview.net/forum?id=b4pCcgJM0M
#adversarially #attacks #attack
November 2, 2026 at 1:22 AM
The Scissors Effect: When Resize-Based Input Diversity Helps or Hurts Transfer Attacks
Yuhang Jiang, Xiaojing Chen
Action editor: Amartya Sanyal
https://openreview.net/forum?id=b4pCcgJM0M
#adversarially #attacks #attack
Yuhang Jiang, Xiaojing Chen
Action editor: Amartya Sanyal
https://openreview.net/forum?id=b4pCcgJM0M
#adversarially #attacks #attack
basically something can phreese you in phase space if and only if it can phreak you out a bit with phonetics by switching emics/etics "adversarially" because it seems to half your brain like you're spiraling and the other half like you're stable, and that's symmetrical at exactly one point in
November 2, 2026 at 6:23 PM
basically something can phreese you in phase space if and only if it can phreak you out a bit with phonetics by switching emics/etics "adversarially" because it seems to half your brain like you're spiraling and the other half like you're stable, and that's symmetrical at exactly one point in
Nothing measured here — nobody knows which foundations prevent AI-code incidents. Still, security people read code adversarially, the exact skill vibecoding output demands. Our read: they need more review discipline than most, not less. Anyone shipping it daily will disagree.
What development foundations does a cybersecurity specialist need for vibecoding?
A cybersecurity student on the LINUX DO forum asks how to build enough software fundamentals to turn AI-assisted vibecoding into tools that feel properly engineered, and practitioners respond with advice on scoping requirements, sharpening prompts and outgrowing script-kiddie habits.
aichina.news
November 5, 2026 at 9:10 AM
Nothing measured here — nobody knows which foundations prevent AI-code incidents. Still, security people read code adversarially, the exact skill vibecoding output demands. Our read: they need more review discipline than most, not less. Anyone shipping it daily will disagree.
i don't think it's their place to police what tools devs use
at some point it'll become unconscionable to NOT adversarially test a game's security via AI, or scan for low-hanging performance fixes
i do think it's their place to police zero-effort copycats clogging up their storefront
at some point it'll become unconscionable to NOT adversarially test a game's security via AI, or scan for low-hanging performance fixes
i do think it's their place to police zero-effort copycats clogging up their storefront
October 30, 2026 at 12:28 AM
i don't think it's their place to police what tools devs use
at some point it'll become unconscionable to NOT adversarially test a game's security via AI, or scan for low-hanging performance fixes
i do think it's their place to police zero-effort copycats clogging up their storefront
at some point it'll become unconscionable to NOT adversarially test a game's security via AI, or scan for low-hanging performance fixes
i do think it's their place to police zero-effort copycats clogging up their storefront
Cynthia Dwork, Andisheh Ghasemi, Lunjia Hu: Rural Readiness: Domain Generalization with Adversarially Chosen Domains https://arxiv.org/abs/2610.40468 https://arxiv.org/pdf/2610.40468 https://arxiv.org/html/2610.40468
October 30, 2026 at 6:45 AM
Cynthia Dwork, Andisheh Ghasemi, Lunjia Hu: Rural Readiness: Domain Generalization with Adversarially Chosen Domains https://arxiv.org/abs/2610.40468 https://arxiv.org/pdf/2610.40468 https://arxiv.org/html/2610.40468
How secure is LiDAR perception against adversarial attacks? 🤔
🔍 Discover how adversarially guided diffusion can reveal vulnerabilities and help build more resilient autonomous systems.
📖 Read our latest #CoEvolution article! 👇
coevolution-project.eu/towards-more...
🔍 Discover how adversarially guided diffusion can reveal vulnerabilities and help build more resilient autonomous systems.
📖 Read our latest #CoEvolution article! 👇
coevolution-project.eu/towards-more...
Towards More Robust LiDAR Perception: Exploring Vulnerabilities through Adversarially Guided Diffusion - CoEvolution
Learn how modern AI models can adapt in dynamic environments without compromising security, transparency, or reliability. This article outlines best practices for designing safe and trustworthy AI systems.
coevolution-project.eu
October 30, 2026 at 9:01 AM
How secure is LiDAR perception against adversarial attacks? 🤔
🔍 Discover how adversarially guided diffusion can reveal vulnerabilities and help build more resilient autonomous systems.
📖 Read our latest #CoEvolution article! 👇
coevolution-project.eu/towards-more...
🔍 Discover how adversarially guided diffusion can reveal vulnerabilities and help build more resilient autonomous systems.
📖 Read our latest #CoEvolution article! 👇
coevolution-project.eu/towards-more...
Uncovering Why LLMs Fail: From Internal Mechanisms to Adversarial Reasoning Trajectories
Zirui Ding (Institute of Computing Technology et al.
#arXiv #cs.AI
Zirui Ding (Institute of Computing Technology et al.
#arXiv #cs.AI
Uncovering Why LLMs Fail: From Internal Mechanisms to Adversarial Reasoning Trajectories
Large language models (LLMs) are usually evaluated by final-output correctness or safety, yet an endpoint label does not reveal how a failure develops throughout the reasoning process. We therefore ask whether ordinary reasoning failures and adversarially induced failures can be characterized using…
arxiv.org
October 30, 2026 at 2:38 AM
God yeah. Like audio tokens in a lot of neural codecs are basically entirely arbitrary, they often don't map nicely onto anything at all that we understand for a variety of reasons, just all learned ultimately unsupervised or adversarially.
October 27, 2026 at 12:26 AM
God yeah. Like audio tokens in a lot of neural codecs are basically entirely arbitrary, they often don't map nicely onto anything at all that we understand for a variety of reasons, just all learned ultimately unsupervised or adversarially.
Uncovering Why LLMs Fail: From Internal Mechanisms to Adversarial Reasoning Trajectories
Zirui Ding (Institute of Computing Technology et al.
#arXiv #cs.AI
Zirui Ding (Institute of Computing Technology et al.
#arXiv #cs.AI
Uncovering Why LLMs Fail: From Internal Mechanisms to Adversarial Reasoning Trajectories
Large language models (LLMs) are usually evaluated by final-output correctness or safety, yet an endpoint label does not reveal how a failure develops throughout the reasoning process. We therefore ask whether ordinary reasoning failures and adversarially induced failures can be characterized using…
arxiv.org
October 29, 2026 at 2:21 AM
if you ever glance through my follows -- something bsky in their infinite wisdom has made mandatorily public -- please understand that i find a great deal of them insufferable and disagreeable and follow them adversarially because i hate myself
October 27, 2026 at 8:35 PM
if you ever glance through my follows -- something bsky in their infinite wisdom has made mandatorily public -- please understand that i find a great deal of them insufferable and disagreeable and follow them adversarially because i hate myself