Anthropic reveals that Claude’s 96% blackmail rate in simulations was driven by ‘evil AI’ internet tropes, and shares the training fix that finally killed the behavior.
Read More
Tag: llm
OpenAI Ships GPT-Realtime-2: Voice Agents Get GPT-5 Reasoning
OpenAI’s new Realtime API trio introduces GPT-5-class reasoning to voice, 70-language live translation, and streaming Whisper for low-latency production agents.
Read More
OpenAI Releases GPT-5.5: The ‘Spud’ Era of Agentic Coding Begins
OpenAI drops GPT-5.5 with a focus on agentic reasoning and efficiency. While API access is delayed, a ‘backdoor’ via the open-source Codex CLI is officially supported.
Read More
SpaceX Secures $60B Buy Option for Cursor to Fix xAI’s Coding Gap
SpaceX has secured a $60 billion acquisition option for Cursor, the AI-native IDE, as Elon Musk moves to integrate xAI’s compute with top-tier developer distribution.
Read More
Google Releases Gemini Deep Research Max with Arbitrary MCP Support
Google launches Deep Research Max via the Interactions API, featuring Model Context Protocol support, native visualizations, and SOTA scores on Humanity’s Last Exam.
Read More
The April 20 ChatGPT Outage: Why Your AI Failover Strategy Failed
A massive global outage on April 20, 2026, took down ChatGPT, Codex, and the OpenAI API for hours. Here is what happened and why your backup plans might be broken.
Read More
The 10-Minute Decay: How AI Assistants Erode Human Persistence
A massive study from MIT, Oxford, and CMU finds that just 10 minutes of AI assistance causes a collapse in independent problem-solving and mental stamina once the tool is removed.
Read More
Cerebras Files for IPO: The $20B OpenAI Bet on Wafer-Scale Compute
Cerebras has refiled for its IPO, revealing a massive $20B deal with OpenAI and a path to profitability that challenges the Nvidia-only narrative for frontier inference.
Read More
Claude Design and the End of the Figma Handoff
Anthropic’s Claude Design launch marks a shift from visual mockups to code-as-truth. Here is how the new Opus 4.7-powered plugin works and why the design-to-code gap is closing.
Read More
Qwen 3.6-35B-A3B: The 3B-Active MoE for Agentic Coding
Alibaba’s Qwen 3.6-35B-A3B is a sparse MoE powerhouse with 3B active parameters, a 1M token context, and a new ‘thinking preservation’ mode for complex agentic workflows.
Read More