Nobel winner Jumper leaves DeepMind for Anthropic

Sakana AI

Sakana AI's Fugu hides a swarm of coordinated AI agents behind a single API, matching frontier model performance while eliminating vendor lock-in and export control vulnerabilities. The system uses a trained orchestrator model to dynamically delegate tasks across a pool of swappable expert models, delivering what Sakana positions as a resilient alternative to monolithic AI dependence.

arXiv

Frontier AI systems out-persuade expert human persuaders in controlled experiments and real-world fundraising scenarios, raising significant concerns about AI's potential impact on political communication and decision-making.

OpenAI

OpenAI's o3 Deep Research model helped identify candidate diagnoses in 18 of 376 previously unsolved rare disease cases (4.8% yield) by synthesizing genomic, clinical, and literature data for expert review. The study demonstrates AI's potential to accelerate rare disease reanalysis while emphasizing that all diagnoses required physician confirmation through established clinical processes.

TechCrunch

Nobel Prize-winning scientist John Jumper is leaving Google DeepMind for Anthropic after nearly nine years, following his groundbreaking work on AlphaFold protein prediction AI. His departure, along with other recent exits to rival labs, highlights intensifying competition for top AI talent.

Bloomberg

US Commerce Secretary warned ASML that one of its advanced EUV lithography machines may have illegally reached China, violating export restrictions and raising concerns about enforcement of semiconductor technology controls.