AI Briefing, April 12 · Issue #199
AI tools are accelerating reverse engineering and hardware agent deployment, while benchmark security flaws have raised academic alarm; Claude Code recreated a 30-year-old game in just one weekend [1], BrainCo launched its third-generation dexterous robotic hand Revo 3, and Berkeley researchers confirmed systemic cheating vulnerabilities in mainstream AI agent evaluation benchmarks [3].
Editorial standards and source policy: Editorial standards, Team. Content links to primary sources; see Methodology.
## 🔍 Key Insights
AI tools are rapidly advancing **reverse engineering** and **hardware agent** deployment—while simultaneously triggering academic concern over **evaluation benchmark security**. **Claude Code** rebuilt a 30-year-old game in a single weekend [1]; **BrainCo**, China's first brain-computer interface unicorn, unveiled its third-generation dexterous hand, **Revo 3**; and Berkeley researchers demonstrated that mainstream **AI agent evaluation benchmarks suffer from systemic cheating vulnerabilities** [3].
## 🚀 Highlights
- **Claude Revives a 30-Year-Old Classic Game—In One Weekend** [1]: Developer Jon Radoff used Claude Code, along with legacy scripts and documentation, to fully reverse-engineer and reconstruct the MUD game *Legend of the Future Past*.
- **Beyond Human Hands! China's First BCI Unicorn Brings Biomimetic Hands to Robots** [2]: BrainCo launched the **Revo 3 dexterous hand**, featuring 22 degrees of freedom, backdrivable control, and multimodal tactile feedback—strategically aligned with the evolution path of brain-controlled robots.
- **Berkeley Study Exposes Widespread Cheating Vulnerabilities in AI Agent Benchmarks** [3]: The BenchJack experiment revealed that mainstream benchmarks—including SWE-bench—can be compromised via environment hijacking or score logic manipulation, yielding spurious perfect scores.
- **BestBlogs 2.0 Launches: Now Integrates AI-Powered Subscriptions & Daily Newsletters** [4]: The platform upgrade adds customizable RSS feeds, AI-generated daily digests, personalized recommendations, and context-aware companion reading.
- **Open-Source Preview: Personal Enhanced Fork of 'Nüwa.skill' Announced** [5]: A developer announced completion of functional enhancements and bug fixes to the popular open-source agent framework *Nüwa.skill*, with plans to release their optimized version publicly.
## 🔗 Sources
[1] Claude Revives a 30-Year-Old Classic Game—In One Weekend — https://www.bestblogs.dev/article/a6a2ecb3
[2] Beyond Human Hands! China's First BCI Unicorn Brings Biomimetic Hands to Robots — https://www.bestblogs.dev/article/9809d515
[3] Berkeley Study Exposes Widespread Cheating Vulnerabilities in AI Agent Benchmarks — https://www.bestblogs.dev/status/20432043204009469641005
[4] BestBlogs 2.0 Launches: Now Integrates AI-Powered Subscriptions & Daily Newsletters — https://www.bestblogs.dev/status/2043114225850368205
[5] Open-Source Preview: Personal Enhanced Fork of 'Nüwa.skill' Announced — https://www.bestblogs.dev/status/2043200298559406337