← The Polylog AI Intelligence Brief
Morning Edition · Thursday, June 18, 2026
TIGER Recovers Training Inputs From Federated Learning Gradients
A new attack reconstructs private inputs by optimizing distances in embedding subspaces, sharpening doubts about gradient sharing.
A paper posted to arXiv presents TIGER, a gradient inversion attack that reconstructs private inputs from the gradient updates clients send in federated learning. The method works by inverting transformer gradients through optimization over…
Continue the AI Intelligence Brief
Track frontier labs, chips, export controls, model releases, regulation, and AI infrastructure.
- 5 AI intelligence signals a day
- Frontier labs, compute, and chips
- Model releases and AI infrastructure
- Source-grounded analysis with confidence labels
The Global Intelligence Brief stays free.
More from this edition
- U.S. Suspends Access to Anthropic's Fable 5 and Mythos 5 Under Export Control Directive
- Google Reports AMIE Matched Primary Care Physicians on Disease Management in Nature
- OpenAI Says a Near-Autonomous AI Chemist Built on GPT-5.4 Improved a Drug Reaction
- DeepSeek Adds Vision to Its Assistant
- Alibaba's Qwen Releases a Three-Model Robotics Suite for Embodied AI
- Moonshot Releases a High-Speed Mode for Kimi K2.7 Code
- ChatGPT's Share of the AI Assistant Market Falls Below 50 Percent, Sensor Tower Says
- France Brings Sovereign AI Compute Online With NVIDIA Hardware
- OpenAI Details a Deployment Simulation Method to Catch Model Failures Before Release
- OpenAI Introduces LifeSciBench for Real-World Life Science Tasks
- Gaussian Mixture Attention Proposes Linear-Time Sequence Mixing for Long Contexts