Get the app

Today's Bytes

The rest of the day's AI news, briefly. One page per day. Page 1 of 10.

6 items

November 11, 2026

Alibaba opens Qwen4-Max preview access on Token Plan for agent developers · Microsoft shows editing ten dormant neurons installs invisible backdoors in agent models · Expo releases Exact to let coding agents build and test native cross-platform apps

6 items

November 10, 2026

FP4 pretraining requires up to 50% more exposures to memorize rare facts · WorkflowArena benchmarks autonomous workflow orchestration across multi-agent language model systems · KiteNorm stabilizes deep Post-LayerNorm transformers via variance regularization and residual scaling

6 items

November 9, 2026

Generalist AI models surpass specialist policies in zero-shot robot manipulation benchmarks · BitDecoding-FP4 enables native FP4 KV cache decoding with near-BF16 accuracy · Parallel agent teams scale test-time compute better than large individual teams

6 items

November 6, 2026

Researchers introduce VFlash to speed up robot visual trajectory planning by 25x · Stanford audit finds maintenance defects in thirty percent of agentic evaluation benchmarks · Hybrid State Transfer enables twenty-fold faster context handoff across linear-attention models

6 items

November 5, 2026

Online reinforcement learning enables multi-task adaptation in generalist robot foundation policies · Systolic linear attention accelerates gated delta attention training on Google TPUs · SpecCompact achieves up to fivefold speedups in language model context compaction

6 items

November 4, 2026

NVIDIA researchers find semantic compaction eliminates need for massive context windows · Princeton and UIUC introduce PEARL to durably suppress model misalignment during post-training · Stanford and NVIDIA study shows gradient coupling improves robot manipulation and driving

7 items

November 3, 2026

Anthropic amended S-1A warns AI progress may soon trigger recursive self-improvement · Honeycomb launches goal-oriented AI SRE agents to monitor production releases · UK cyber agency warns AI ransomware will likely hit infrastructure by 2028

6 items

November 2, 2026

Implicit Dynamics World Models decouple state reachability from low-level robotic control · Latent Sparse Attention accelerates ten-million-token context processing by up to 72x · Flide inference engine accelerates agent decision model branching up to ninefold

5 items

November 1, 2026

Elon Musk details 420,000 GB300 GPU clusters for xAI training and inference · Five Qwen3.8 agents outperform Claude Fable 5 on LiveCodeBench Hard · Microsoft enforceable school AI guardrails ban student data training nationwide

6 items

October 31, 2026

Alphabet and Microsoft report Q3 2026 earnings driven by AI infrastructure · Dodaios releases open-source operating system designed for multi-agent orchestration · Signet AI launches cryptographic framework for verifiable LLM state transitions

6 items

October 30, 2026

VeriCUA introduces agentic exploration for computer-using agent verification · SwapBot enables modular robotic tool use with tactile interchangeable adapters · Multi-View MoE modulates sparse expert routing without additional compute

4 items

October 29, 2026

OpenAI gives ChatGPT Pro and Business subscribers up to $500 monthly API credits · Meta introduces SixthView benchmark to evaluate 3D spatial reasoning in multimodal models · Samsung reports record quarterly operating profit as HBM4 memory sales triple