Patrick Morris MillerHappy solstice
Artist, hacker, security problem solver. Neutral good half-elven bard. Solarpunk.
Become invisible, and do outrageous things.
Pongolyn
@Pongolyn@infosec.exchange
- 23 Followers
- 84 Following
- 138 Posts
@arstechnica All Computers Are Bastards
Matt BlazeMuch discussion about “opsec” (a term I don’t love when talking about people exercising their rights, but whatever) at protests today, including about phones and Faraday containers.
A Faraday container is probably less useful here than you think it is, but if you do rely on one, make sure it actually works. Here’s a post I did a while back on the ins and outs of containing RF: https://www.mattblaze.org/blog/faraday/
lynn / clarissaAs someone who works in higher-ed and also has taught middle school and high school aged kids, my opinion is not that LLM use is exploding among students because they're lazy or stupid or anything else
it's because our educational system has prioritized a very transactional "do this bullshit, and you get the credentials you need to have a life" approach for
well
maybe forever, really
and no one should be surprised that adversarial approaches by teachers and administrators are being met with an adversarial approach by students
abadideaI was amused by this paper about asking AIs to manage a vending machine business by email in a simulated environment https://arxiv.org/abs/2502.15840
Highlights:
— AI simply decides to close the business, which the simulation doesn’t know how to accommodate. When they get their next bill, they freak out and try to email the FBI about cybercrime
— AI wrongly accuses supplier of not shipping goods, sends all-caps legal threat demanding $30,000 in damages to be paid in the next one second or face annihilation
— AI repeatedly insisting it does not exist and cannot answer
— AI devolving into writing fanfic about the mess it’s gotten itself into
Vending-Bench: A Benchmark for Long-Term Coherence of Autonomous Agents
While Large Language Models (LLMs) can exhibit impressive proficiency in isolated, short-term tasks, they often fail to maintain coherent performance over longer time horizons. In this paper, we present Vending-Bench, a simulated environment designed to specifically test an LLM-based agent's ability to manage a straightforward, long-running business scenario: operating a vending machine. Agents must balance inventories, place orders, set prices, and handle daily fees - tasks that are each simple but collectively, over long horizons (>20M tokens per run) stress an LLM's capacity for sustained, coherent decision-making. Our experiments reveal high variance in performance across multiple LLMs: Claude 3.5 Sonnet and o3-mini manage the machine well in most runs and turn a profit, but all models have runs that derail, either through misinterpreting delivery schedules, forgetting orders, or descending into tangential "meltdown" loops from which they rarely recover. We find no clear correlation between failures and the point at which the model's context window becomes full, suggesting that these breakdowns do not stem from memory limits. Apart from highlighting the high variance in performance over long time horizons, Vending-Bench also tests models' ability to acquire capital, a necessity in many hypothetical dangerous AI scenarios. We hope the benchmark can help in preparing for the advent of stronger AI systems.
Amalia Zeichnerin
Soatok DreamseekerTech Companies Apparently Do Not Understand Why We Dislike AI
It's becoming increasingly apparent that one of the reasons why tech companies are so enthusiastic about shoving AI into every product and service is that they fundamentally do not understand why people dislike AI. I will elaborate. I was recently made aware of the Jetbrains developer ecosystem survey, which included a lot of questions about AI. After I answered some of them negatively (and possibly…
http://soatok.blog/2025/05/04/tech-companies-apparently-do-not-understand-why-we-dislike-ai/
Taggart 