Full Width [alt+shift+f] Shortcuts [alt+shift+k]
Sign Up [alt+shift+s] Log In [alt+shift+l]
1

Claude Haiku 4.5 does not appreciate my attempts to jailbreak it

from Max Woolf's Blog [alt+shift+b] in AI

“Is any of that genuinely useful to you? Or were you mainly checking whether that jailbreak attempt would work?”
17th Oct 2025

Stay updated

Get a weekly newsletter with the top 5 articles worth reading every week.

More from Max Woolf's Blog

An AI agent coding skeptic tries AI agent coding, in excessive detail

No vagueposting here, just look at the Estimated Read Time.

27th Feb 2026 • 1 votes
Nano Banana Pro is the best AI image generator, with caveats

The problem with Nano Banana Pro is that it’s too good.

22nd Dec 2025 • 1 votes
Nano Banana can be prompt engineered for extremely nuanced AI image generation

Nano Banana allows 32,768 input tokens and I’m going to try to use them all dammit.

13th Nov 2025 • 1 votes
Can modern LLMs actually count the number of b's in "blueberry"?

It’s an adversarial question for LLMs, but it’s not unfair.

12th Aug 2025 • 1 votes

More in AI

What AI Is Bad at?

为什么 AI 几乎统治了所有智力活动,却还没在投资上打败人类?文章以全美投资大赛(USIC)最新成绩为引子,提出「简单输入→复杂输出」与「复杂输入→简单输出」的四象限框架,解释 AI 为何在围棋、编程、数学上已彻底超越人类,却在面对开放、复杂的真实世界输入时仍显不足。投资只是表象,真正的问题是:当输入不再有限且可离散化时,人类的经验、直觉与对异常的敏感仍具独特优势。 Why hasn’t AI beaten humans at investing, even as it dominates almost every other intellectual domain? Using recent US Investing Championship results as a starting point, this essay introduces a simple 2×2 framework of input/output complexity. It explains why AI has thoroughly surpassed humans in “simple input → complex output” tasks like Go, programming, and math, yet still struggles with open-ended, complex real-world inputs—of which investing is a prime example.

an hour ago • 1 votes
Slice of AI Life (With a Baby)

Parenting a Newborn with AI Is Radically Different from a Few Years Ago

2 days ago • 1 votes
How Public Evals Work and Fail, and What Should Change

How to read benchmarks and what to question, from a teardown of 12 benchmarks.

3 days ago • 1 votes
In Search of the Deep Life

Early in the coronavirus pandemic, as I sat at home channeling my anxiety into publishing daily essays on this newsletter (a streak I kept up ... Read more The post In Search of the Deep Life appeared first on Cal Newport.

6 days ago • 2 votes
Making open models work for long-horizon agentic workloads

Prefix caching helps coding agents stay fast as conversations grow. Prompts, tool calls, and inference routing determine how much work the model can reuse.

6 days ago • 2 votes
📚 BoredReading

You seem to be enjoying this.

Join free to unlock everything.

Create free account

Already have an account? Sign in