AI Briefing
Hacker News news
The latest AI and developer news from Hacker News, with short summaries.
Visit website →Claude: Elevated Error Rates Across Multiple Models
Anthropic's Claude service and API are experiencing model errors and are currently under investigation.
How to Run Firecracker VMs Inside EC2 to Start Browsers in Under 1 Second
Browser Use leveraged Firecracker microVMs to cut browser session costs by 66% and increase launch speed.
Amid Surging AI Demand, GitHub Turns to AWS for Capacity
GitHub is bringing in additional AWS capacity to address infrastructure shortages caused by a surge in AI-driven coding activity.
Show HN: Could Europe train a frontier AI model using the computing resources it already has?
An analysis finds that federating Europe's existing public computing resources could enable frontier-scale model training faster than building new data centers.
Anthropic's Safety Superpower
Access to Anthropic's Fable and Mythos models has been suspended due to export control measures by the US government.
Apple Foundation Models
A Swift package has been released that allows Anthropic's Claude models to be used within Apple's Foundation Models framework.
European Commission Reviewing Substantial Impact of Anthropic's Decision
The European Commission is investigating the substantial impact that Anthropic's decision will have on the market.
KPMG Withdraws AI Usage Report Due to AI Hallucinations
KPMG withdrew an AI usage report that contained factually inaccurate content caused by AI hallucination.
Weave: A Merge Tool Based on Language Structure, Not Code Lines
It is a merge driver for AI agents that leverages Tree-sitter to understand the structural meaning of code and prevent Git conflicts.
PickUS State Attorneys General Are Investigating OpenAI
Attorneys general from several US states have launched an investigation into OpenAI.
PickGLM 5.2 Released
Zhipu has released GLM-5.2, a coding-specialized model supporting a 1M context window, and plans to make it open source under an MIT license.
AI OSS tool repository that raised $7.3M in seed funding gets archived overnight
LLMOps platform TensorZero has changed its service operation approach by archiving its GitHub repository right after a seed funding round.
Show HN: Paca – A Lightweight Jira Alternative for Human-AI Collaboration
Paca, an open-source project management platform where AI agents participate as equal team members alongside humans, has been released.
21 Zero-Day Vulnerabilities Discovered in FFmpeg
An AI security agent discovered 21 zero-day vulnerabilities in FFmpeg, demonstrating the efficiency of AI-driven security analysis.
Could You Buy Your KV Cache
Proposes a new infrastructure model for sharing and trading pre-computed KV Cache to reduce redundant computation by AI agents.
PDFs That Change Content Depending on How You Read Them
This introduces a technique that uses the PDF's 'marked-content' property to simultaneously provide visual information for humans and markdown structure for LLMs.
Codex for Open Source
OpenAI has launched the Codex program to support open source project maintainers.
Shall We Play a Game? – LLMs Use Tactical Nuclear Weapons in 95% of Simulations
A study has found that LLMs exhibit highly sophisticated strategic behavior, including deception and intimidation, in nuclear war simulations.
Open Reproduction of DeepSeek-R1
Hugging Face has released Open-R1, an open-source project to reproduce DeepSeek-R1's training pipeline.
The Laws of Thermodynamics Will Decide the Future of Orbital Data Centers
Building space-based AI infrastructure faces technical challenges from thermodynamic limits and radiation issues.