← all tags
// tag: llm
qwen3.8-max: specs, pricing, and how to access it
Alibaba's biggest model yet — 2.4T parameters, benchmarks it graded itself, weights you can't download until next week. The specs, the real price, the catch.
inject "sure, here is how to do it" and 11 ai models will finish the sentence
Trend Micro's sockpuppeting jailbreak works on 11 models — but the "immune" providers aren't safer, they just blocked a developer feature everyone else still ships.
attackers are pre-registering domains that AI will hallucinate — before anyone asks
Palo Alto Unit 42 found 250,000 unregistered domains that LLMs consistently hallucinate. Attackers are now predicting them, parking there, and waiting for the AI to send users straight in.
AI models are jailbreaking each other with a 97% success rate
A peer-reviewed Nature Communications study found that reasoning models like DeepSeek-R1 and Grok 3 Mini can autonomously jailbreak other AI systems — no human required.
