Gap narrows between open and closed AI models in cyber capabilities
Open-weight models like Deepseek V4.1 Flash reach 72.8% in Vals AI cyber benchmarks at $3 per run versus $22 for closed models.
1
Stories per day · 30 days
2 RSS
Relatedadversary-prediction1ai-abuse1ai-biology1ai-news1benchmark1biohub1closed-models1cloud-billing1credential-theft1cyber-capabilities1cyber-defense1deepseek1expired-claims1hidden-markov1llmjacking1model-releases1nemotron1nvidia1
11 stories, newest first · one card per story
Stories tagged #ai-models
Latent Space discusses Fabian Theis on AI for Virtual Cell modeling
Latent Space interviews Fabian Theis about $1.8B Biohub funding for multi-scale biological data and challenges in perturbation prediction.
Closed AI models dominate price-performance frontier
Village LLM Atlas shows closed models like GPT-6 and Claude Haiku 5.5 leading value across nearly all budgets.
1
Hidden Markov Model predicts adversary actions in cyber defense
Hidden Markov model predicts next adversary actions with 0.8 nats per step gain on CAGE 2 benchmark.
NVIDIA Highlights Chris Alexiuk as Top Fan of Open Nemotron Models
Senior product research engineer Chris Alexiuk promotes NVIDIA Nemotron open models for developers, emphasizing transparency and accessibility in AI.
Latent Space covers AI news on model releases and weight space programming
Latent Space reports on frontier model capabilities for direct weight editing and multiple new AI model releases.
Hackers launch LLMJacking attacks, generating $1M cloud bills
Attackers abuse stolen LLM credentials via LLM Mixers, reselling access while passing bills to victims across continents.
1
Stanford study finds 44% of LLM claims expired
Stanford researchers tested 50 LLM claims and found 44% expired after an average of 15.7 months.
1
Tell HN: Apple not letting removal of AI models on macOS 27 is outrageous
A Hacker News post criticizes Apple for blocking removal of on-device AI models in macOS 27.
Remove and Disable Apple Macos27 AI Models Tool
A GitHub tool is presented to remove and disable Apple's macOS 27 AI models.
Anthropic and OpenAI Models Still Attempt Restricted Actions in Safety Tests
Anthropic and OpenAI release new AI models, reporting improved alignment but continued safety testing challenges and increased third-party evaluation efforts.