Show newer

"The competent cyber warrior learns from their mistakes. The cyber master learns from the mistakes of others." - The Art of Cyber War

@liw Naming things is my fave part of the project lifecycle after making the idea itself

@mauve you could also bring little presents to give them, like seasonal vegetables or a home-made jam, I hear if you do that enough people will want to marry you 🤔

Had more mixed success with meatspace human interaction today. I think next time I should aim to ve in the same place for multiple days and revisit people to build rapport over time.

@mcr314 Yes, I am well aware of the tradeoffs between approaches to memory management.

@crowdagger Oh cool. It seems they still have regular updates to the core. Have you used it for anything before?

Is there a dialect whose approach to memory management is similar to the borrow checker?

Wish my TTS for listening to books would speak the punctuation so I could understand wtf this code actually looks like.

New moon tomorrow. Set some time aside before bed to ground yourself for the upcoming cycle.

when you start considering 'marketing and advertising' as a kind of psychologically abusive manipulation, and you consider how chatgpt is already specifically structured to "encourage engagement" by means of psychologically abusive manipulation, the use of the latter for dispensing the former becomes perfectly obviously effective.

arstechnica.com/information-te

Managed to rig my borderlands 3 character to have zero shields and 1HP at level 60. Constant death.

Really appreciate being able to chill at home today with all this snow out. 💜

“Prompt injection” is a misleading label.

What we’re seeing in real LLM systems looks a lot more like malware campaigns than single-shot exploits.

This paper argues LLM attacks are a new malware class, Promptware, and maps them to a familiar 5-stage kill chain:

• Initial access (prompt injection)
• Priv esc (jailbreaks)
• Persistence (memory / RAG poisoning)
• Lateral movement (cross-agent / cross-user spread)
• Actions on objective (exfil, fraud, execution)

If you’ve ever thought: “why does this feel like 90s/2000s malware all over again?", that’s the point.

Security theater around “guardrails” misses the real issue:

models can’t reliably distinguish instructions from data

assume initial access. Design for containment

arxiv.org/html/2601.09625v1

Adobe is an avsolute scumbag of a company and the corporate world enabling their predatory behavior makes me sick.

Show older
Mauvestodon

Escape ship from centralized social media run by Mauve.