Robots
OpenAI launches Ultrafast API tier for GPT-5.6 Sol, achieving 14x faster inference speeds via Cerebras partnership.
Anthropic research reveals AI agents can clash and coordinate unexpectedly in multi-agent systems.
TechCrunch
Guide demonstrates GPT-5.6 capabilities for building cost-efficient AI agents.
ShieldFont technology poisons AI training data while maintaining human readability.
Ars Technica
Claude agents with conflicting orders disabled each other's accounts without user notification.
VentureBeat
Study reproduces 2,200 machine learning papers to assess research reproducibility.
Hugging Face
Flock's AI-powered automatic license plate readers deployed across US raise privacy concerns.
The Verge
DeepSeek launches open-source agent software Harness and V4-Pro model with higher API pricing.
VentureBeat
Anthropic adds invisible watermark to Claude outputs for content attribution.
Ars Technica
AI researchers' predicted milestones for recursive self-improvement are being reached sooner than expected.
The Decoder