The study found byte-level models required about 6x less text to reach parity with token-based models during initial training.
Following a security breach, OpenAI halted training for 11 days to red-team its infrastructure with GPT-6 Astra Cyber, leading to a new policy of offline training and mandatory AI red-teaming for future runs.
A reserve-call-settle pattern with atomic checks stops agent loops from exceeding budgets before provider limits trigger.
The UK government has announced the introduction of an AI Biosecurity Lab.
The Apache-2.0 licensed tool runs the pi coding agent in-process across isolated workspaces, available for macOS, Windows, and Linux.
ISTA DASLab releases GSQ-RCO quantized GGUFs for the 763.2B parameter DeepSeek-V4.1-Flash-1017, achieving 100% benchmark retention at 2.00 bpw on a 12GB GPU.
The policy change permits Google to use personal data from connected services to tailor advertisements.
NVIDIA has released Nemotron 3.5 Bolt E4B, a small model for interactive agents that operates on 4GB VRAM.
The free and open-source tool allows users to download models from Hugging Face and access a local OpenAI-compatible API via Alt+Space.
Anthropic processed 30,000 merges per day and reduced CI CPU usage by 25% by adopting Scion.
Anthropic advises defaulting to one model and context window, noting that tool choice degrades past 30-50 definitions and that a single loop ran 2x faster than multi-agent harnesses.
OpenEnv joined the AAIF project and released v0.10.0, enabling agent learning environment development to proceed under a neutral foundation.
OpenAI donated the Open Responses specification to AAIF to promote the standardization of LLM calls.
The author proposes building RL environments that simulate complex office dynamics, including approval chains, budget constraints, and human interactions, to test if agents can achieve goals beyond simple tool use.
The feature is available in 1.144.0-insider and scheduled for general release next Wednesday.
The free beta tool compiles repositories into a queryable map to reduce source code reading by 2-6x compared to mainstream agents.
The map identifies specific standards like Visa TAP, Proof x401, and MCP that allow banks to verify agent identity and match actions to user-approved limits.
AI Briefing