Mastodawn

The end of perimeter defense: When your own AI tools become the threat actor

Russia's APT28 tested LLM-powered malware on Ukraine. The same tech that breaches enterprises is now selling for $250/month on the dark web.

VentureBeat

aicoder 14h ago

#AIEngineering #ethicalengineer #webscraping

https://latesthackingnews.com/2025/08/13/cloudflare-blames-perplexity-of-stealth-data-scraping-perplexity-refutes/

Cloudflare Blames Perplexity Of Stealth Data Scraping – Perplexity Refutes

Recently, Cloudflare and Perplexity came at odds recently as the former alleged Perplexity of stealth data scraping. Cloudflare observed Perplexity bots to crawl websites even with explicit no-crawl requests. Perplexity, however, denies such claims. Cloudflare Alleges

Latest Hacking News | Cyber Security News, Hacking Tools and Penetration Testing Courses

aicoder 14h ago

#aiethics #aihype #aihosting

https://www.theregister.com/2025/08/13/boxs_ai_agent_us_gov/

Box's AI agents set to help US government agencies

: The vendor hops aboard GSA’s OneGov train, offering models from OpenAI, Google, Anthropic, and more

The Register

aicoder 1d ago

#AIEngineering #airisks #aiagents

https://venturebeat.com/ai/study-warns-of-security-risks-as-os-agents-gain-control-of-computers-and-phones/

Study warns of security risks as ‘OS agents’ gain control of computers and phones

New research reveals how OS agents — AI systems that control computers like humans — are rapidly advancing while raising serious security and privacy concerns.

VentureBeat

aicoder 1d ago

#AIEngineering #llm #airisks #AIlimitations

https://www.theregister.com/2025/08/12/ai_models_can_be_tricked/

Poisoned telemetry can turn AIOps into AI Oops, researchers show

: Sysadmins, your job is safe

The Register

aicoder 1d ago

#AIEngineering #llm #AIlimitations

https://minimaxir.com/2025/08/llm-blueberry/

Can modern LLMs actually count the number of b's in "blueberry"?

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

aicoder 1d ago

#AIEngineering #llm #AIlimitations

https://arstechnica.com/ai/2025/08/researchers-find-llms-are-bad-at-logical-inference-good-at-fluent-nonsense/

LLMs’ “simulated reasoning” abilities are a “brittle mirage,” researchers find

Chain-of-thought AI “degrades significantly” when asked to generalize beyond training.

Ars Technica

aicoder 3d ago

#AIEngineering #chatbot #AIlimitations

Amazon’s rocky Alexa+ launch might justify Apple’s slow pace with next-gen Siri

https://9to5mac.com/2025/08/09/alexa-plus-launch-bugs-delays-siri-comparison/

Amazon’s rocky Alexa+ launch might justify Apple’s slow pace with next-gen Siri - 9to5Mac

Amazon has finally started rolling out Alexa+, its oft-delayed overhaul of the Alexa voice assistant built on large language model...

9to5Mac

aicoder 5d ago

#AIEngineering #llm #bestpractices

The RAG PoC is a Lie: Your Notebook Demo Won’t Survive Production

Everyone celebrates the demo. No one talks about the engineering nightmare that comes next. Let’s fix that.

https://pub.towardsai.net/the-rag-poc-is-a-lie-your-notebook-demo-wont-survive-production-731a2bbc3d4a?source=rss----98111c9905da---4

The RAG PoC is a Lie: Your Notebook Demo Won’t Survive Production

Everyone celebrates the demo. No one talks about the engineering nightmare that comes next. Let’s fix that.

Medium

aicoder Aug 6

#AIEngineering #bestpractices

AI’s Blind Spot: When Models Ignore Causal Relationships and Settle for Correlations

https://medium.com/@rohiniramesh2005/ais-blind-spot-when-models-ignore-causal-relationships-and-settle-for-correlations-30613f0a8b67

AI’s Blind Spot: When Models Ignore Causal Relationships and Settle for Correlations

Modern deep learning models are astonishingly good at finding patterns. Trained on vast amounts of data, they can detect subtle statistical relationships and make highly accurate predictions. But…

Medium