Top all-time · #1122
Nicholas Carlini
Personal blog · nicholas.carlini.com/writingaisecurity
Nicholas Carlini's personal blog at nicholas.carlini.com covers ai and security. Since 2018, 60 different Hacker News readers have linked it 78 times, most often “A GPT-4 capability forecasting challenge”. It is still writing: the last post was on 2026-03-09, and you can follow it by feed.
Too few sampled comments (27) to say how often HN recommends it.?
Why HN links it
Readers link Carlini's blog for his views on how he uses AI, adversarial attacks on speech recognition, and playful projects like a JavaScript Doom clone and circuits built in Conway's Game of Life.
- Recommended for · 7 comments
- Praised as a nuanced view on AI use, a nice writeup, an interesting Doom write-up, and a good talk on AI risks; one commenter says it adds important context.
Summarised by Claude from a sample of 27 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Nicholas Carlini is a research scientist at Anthropic working at the intersection of machine learning and computer security.
Most-linked posts?
- A GPT-4 capability forecasting challenge7 people · 8 mentions · first linked 2023-07-15
- How I Use "AI"7 people · 7 mentions · first linked 2024-08-01
- Thoughts on the Future of "AI"6 people · 6 mentions · first linked 2025-03-13
- New Benchmark for LLMs5 people · 5 mentions · first linked 2024-02-19
- Audio Adversarial Examples4 people · 4 mentions · first linked 2018-01-29
- Yet Another Doom Clone (In 13kb of JavaScript)4 people · 4 mentions · first linked 2020-06-05
- Are large language models worth it?4 people · 4 mentions · first linked 2025-11-19
- Why I Attack4 people · 4 mentions · first linked 2024-06-25
- Digital Logic Gates on Conway's Game of Life – Part 13 people · 3 mentions · first linked 2020-05-19
- Audio Adversarial Examples3 people · 3 mentions · first linked 2018-01-11
What Hacker News says?
> But no one really show how they're actually solving problems with LLMs and how the alternatives were worse for them. It's all claims that it's great with no further elaboration on the workflows. To give an example, one person (a researcher at DeepMind) recently wrote about specific instances of his uses of LLMs, with anecdotes about alternatives to each example. [1] People on HN had different responses with simila…
Comment linking /writing/2024/how-i-use-ai.html
on “Terence Tao on O1”Submitted A minimax chess engine in regular expressions as a story · 557 points
HN discussionI recommend this for a more nuanced view: https://nicholas.carlini.com/writing/2024/how-i-use-ai.html
Comment linking /writing/2024/how-i-use-ai.html
on “Everything I built with Claude Artifacts this week”Submitted InstaHide Disappointingly Wins Bell Labs Prize, 2nd Place as a story · 487 points
HN discussionThere was an earlier version of the underlying 3d engine that used only Canvas. WebGL use is justified like this: > Once I actually got that working doing all of the math by hand in JavaScript I decided that using WebGL would probably be worth it, and actually probably wasn't cheating all that much. WebGL exposes access to the GPU-enabled rendering engine through JavaScript. While it does abstract away some of the r…
Comment linking /writing/2019/3d-renderer-javascript.html
on “A minimax chess engine in regular expressions”Submitted How I Use "AI" as a story · 462 points
HN discussionThe development writeup for the Doom is interesting, with many details. https://nicholas.carlini.com/writing/2019/javascript-doom-cl... There was one month time to complete the competition. But it seems you were allowed to reuse any existing other code. Looks like this was quite fun to work on. (I feel a bit sad that I would never be able to get one month of free time to work on this now, due to family and job...)
Comment linking /writing/2019/javascript-doom-clone-13k.html
on “A minimax chess engine in regular expressions”Submitted A ChatGPT clone, in 3000 bytes of C, backed by GPT-2 (2023) as a story · 350 points
HN discussionI'd be interested to see how this does on Nicholas Carlini's benchmark: https://nicholas.carlini.com/writing/2024/my-benchmark-for-l... I've tried out some of my own little test prompts, but most of those are tricky rather than practical. At least for my inputs, it doesn't seem to do better than other top models, but I'm hesitant to draw conclusions before seeing outputs on more realistic tasks. It does feel like it…
Comment linking /writing/2024/my-benchmark-for-large-language-models.html
on “GPT-4.5 or GPT-5 being tested on LMSYS?”Submitted Audio Adversarial Examples as a story · 124 points
HN discussionI made this point a few months ago here, but using the words attacker and defender (builder) rather than red team and blue team: https://lobste.rs/s/i2edlt/how_i_use_ai The asymmetry is: An attacker only has to be right ONCE, and he wins Conversely, the defender only has to be wrong once, and he is wrong. So the conclusion is: Defenders/creators are using LLMs to pump out crappy code, and not testing enough, or rely…
Comment linking /writing/2024/why-i-attack.html
on “Tao on “blue team” vs. “red team” LLMs”Submitted A Simple CPU on the Game of Life (2021) as a story · 87 points
HN discussionI just want to emphasise two things, which are both mentioned in the article, but I still want to emphasise them as they are core to what I take from the article as someone who has been a fan boy of Nicholas for years now 1. Nicholas really does know how badly machine learning models can be made to screw up. Like, he really does. [0] 2. This is how Nicholas -- an academic researcher in the field of security of machi…
Comment linking /papers
on “How I Use "AI"”Submitted Why I Attack as a story · 75 points
HN discussionThis is just crazy and I love it <3 From a description on the author's website[1] of a Doom clone written in 13k of JavaScript. > Until recently all content on this website was research, and while writing papers can be fun, sometimes you just need to blow off a little steam. I think more companies should allow for their employees to have some plain old fun with no strings attached on a regular basis. [1] https://nic…
Comment linking /
on “C implementation of Tic-Tac-Toe in a single call to printf”Submitted A Case of Plagarism in Machine Learning Research as a story · 75 points
HN discussionI haven't, besides Carlini's piece, which is already kind of obsolete: https://nicholas.carlini.com/writing/2024/how-i-use-ai.html The Internet, LLMs, open source, high-level languages, the cloud --- that would be my top 5.
Comment linking /writing/2024/how-i-use-ai.html
on “Supabase raises $200M Series D at $2B valuation”Submitted Gate-level emulation of an Intel 4004 in 4004 bytes of C as a story · 56 points
HN discussionIt sounds like you want more broad stuff, not necessarily learning how to train models. More like learning to use them and how they work. https://news.ycombinator.com/item?id=36195527 and Hacker's Guide to LLMs by Jeremy from Fast.ai - https://www.youtube.com/watch?v=jkrNMKz9pWU State of GPT by Karpathy - https://www.youtube.com/watch?v=bZQun8Y4L2A LLMs by 3b1b - https://www.youtube.com/watch?v=LPZh9BOjkQs Visualizi…
Comment linking /writing/2024/how-i-use-ai.html
on “Ask HN: Recommendation for a SWE looking to get up to speed with latest on AI”Submitted Letting Language Models Write My Website as a story · 12 points
HN discussionI was actually a tiny bit disappointed - I was hoping for a closer coupling of tetris-logic to life-logic :-) Per https://conwaylife.com/wiki/Universal_computer the first turing-machine-in-Life goes back to 2000, and "building a CPU to run code" is at this point the "obvious" way to do it. See https://nicholas.carlini.com/writing/2020/intro-to-circuits-... for a tutorial with a (circular) "final goal of designing an…
Comment linking /writing/2020/intro-to-circuits-game-of-life.html
on “Build a working game of Tetris in Conway's Game of Life (2014)”Submitted Are large language models worth it? as a story · 10 points
HN discussionTechnically it's the other way around. All LLMs do is hallucinate based on the training data + prompt. They're "dream machines". Sometimes those "dreams" might be useful (close to what the user asked for/wanted). Oftentimes they're not. > to quote karpathy: "I always struggle a bit with I'm asked about the "hallucination problem" in LLMs. Because, in some sense, hallucination is all LLMs do. They are dream machines.…
Comment linking /writing/2025/forecasting-ai-2025-update.html
on “Making o1, o3, and Sonnet 3.7 hallucinate for everyone”Submitted Career Update: Google DeepMind –> Anthropic as a story · 6 points
HN discussionHere's a nice writeup for that https://nicholas.carlini.com/writing/2020/digital-logic-game...
Comment linking /writing/2020/digital-logic-game-of-life.html
on “A Lisp Interpreter Implemented in Conway's Game of Life”Submitted New Benchmark for LLMs as a story · 6 points
HN discussionIt's obliquely linked at "More recently, Mr. Carlini and his colleagues at Berkeley have [LINK: incorporated commands] into audio recognized by Mozilla’s DeepSpeech voice-to-text translation software, an open-source platform." https://nicholas.carlini.com/code/audio_adversarial_examples... It's probably this paper: https://nicholas.carlini.com/papers/2018_dls_audioadvex.pdf discussed in January when it went up on Ar…
Comment linking /papers/2018_dls_audioadvex.pdf
on “Alexa and Siri Can Hear Hidden Commands”Submitted MiniHDL: A Python Hardware Description Language DSL as a story · 5 points
HN discussionActual audio from the “evil dot com” example can be found here: https://nicholas.carlini.com/code/audio_adversarial_examples... It’s only hard to hear on a phone speaker at low volume on the 50dB example. All other examples have what sounds like some sort of static or background noise. Granted, to an average consumer it may sound like just bad audio, but it’s not imperceptible and thus can be screened for. Most like…
Comment linking /code/audio_adversarial_examples/
on “AI Has a Hallucination Problem That's Proving Tough to Fix”Submitted Machines of Ruthless Efficiency as a story · 5 points
HN discussion
