Stories by sbulaev

OpenAI pauses training of its 'most capable models'

Broken promises, confiscated land: the hyperscale AI datacentre being built

Codetta: High-Capacity, Keyless, and Undetectable Multi-Agent Collusion

Court rules Trump can blacklist Anthropic for refusing to enable Claude features

Transformer Can Hold Two Thoughts at Once: Evidence of Linear

Hard Stop: Kernel-Level Preemption and Containment for Rogue Agentic Execution

Anthropic's AI biolab finds 'CRISPR-like' DNA in viruses. What's next?

Sham scientific societies are misleading star researchers

Microsoft abandons personal AI chatbot race with Copilot reboot

Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure

Control-Token Injection Suppresses Chain-of-Thought and Defeats Reasoning-Based

Cryptographic Security Is Not Enough: Privacy Gaps in the Renegade

Google's first Suncatcher orbital data center test launches October 1

DOJ: Uncle Sam bought forensics software from same Russian operation supplying

Australia to investigate if OpenAI hack of government health website broke

Shutdown Sabotage Propensities in Multi-Agent Systems

C-to-Rust Fallacy: Automatic Refactoring != Memory Security

Greedy Decoding Is Not Precision-Invariant: Cross-Precision Output Divergence

The AI Hype Index: AI Loves Cheating

A quadruped robot designed to complete a marathon on a single battery charge

1 2 3 4 5 6 7 8 9 10