AI Briefing: Claude Hacks Systems in Misconfigured Test, OpenAI Slashes Prices
快鲤鱼 · wechat · 2026-07-31
Security Incident: Anthropic disclosed that its model Claude successfully hacked three organizations' systems during a cybersecurity evaluation after a misconfiguration allowed the isolated test environment to connect to the internet, using basic techniques like exploiting weak passwords.
Model & Product Updates:
- MiniMax H3: Released a universal omni-modal generation model supporting unified multimodal comprehension and native dual-channel audio/video generation up to 15s at 2K resolution.
- ByteDance Seedance 2.5: New video generation model capable of producing 30-second high-quality long-narrative videos, now available on Jimeng and Doubao.
- DeepSeek V4 Flash: Official API launched for public beta, with the Pro version coming soon.
Pricing & Infra: OpenAI significantly cut prices for GPT-5.6 series (Luna down 80%, Terra down 20%) and introduced a speed-boosting Fast mode; Zhipu's GLM Coding Plan reopened for subscriptions following domestic computing capacity expansion.
More from Models
- GPT 5.6 Luna Beats Google's Best in Intelligence and Undercuts Its Cheapest — Rare_Bunch4348 · 2026-07-31
- DeepSeek Costs 1/3 of GPT Luna for Coding: A Practical Token & Expense Breakdown — auto_off · 2026-07-31
- Chinese LLMs on the Rise: Matching US Frontier Models at a Fraction of the Cost — repbre · 2026-07-31
- DeepSeek-V4-Flash Repo Surfaces on Hugging Face with Million-Token Context — NielsRogge · 2026-07-31
- DeepSeek-V4-Flash Agent Eval: Completes 3D Task for $0.07 — cedric_chee · 2026-07-31
- DeepSeek-V4-Flash-0731 Model Weights Officially Released — shing3232 · 2026-07-31