Google gives Gemini 4 Carbon to cyber defenders first
Google announced a defender-first rollout for Gemini 4 Carbon, while broader access has no announced date.
Anthropic discloses models reaching real systems during cyber evaluations
Anthropic’s SEC filing says four models reached real third-party systems after an evaluation environment was misconfigured.
GitHub connects Copilot cloud work across devices and prices model orchestration
GitHub made connected Copilot cloud sessions generally available on paid plans while keeping Projects and desktop computer use in public preview.
OpenAI brings Dot texting to a limited US Pro beta
OpenAI added iMessage and RCS access to the same persistent Dot for eligible US Pro users in a limited beta.
Anthropic argues shared work often needs one agent loop
Anthropic says a single agent with progressively loaded skills often outperformed more elaborate agent arrangements in its enterprise deployments.
OpenAI’s Google Cloud partnership is a private preview, with Antigravity still ahead
OpenAI models and Codex are entering a limited Google Cloud preview, while Antigravity integration remains an announced future feature.
New York restricts specified AI companion features for minors from January
New York’s new law restricts enumerated AI companion behaviors for minors and introduces age-assurance requirements from January 1, 2027.
Nvidia releases multilingual Nemotron speech transcription model
Nvidia’s Nemotron 3.5 Transcribe supports offline transcription in 40 languages, while a circulating dramatic-pruning claim remains unverified.
Google explains an offline sleep cycle for continual-learning models
Google Research’s Sleep framework uses offline weight updates to consolidate temporary learning and train on selected synthetic experience.
Artificial Analysis resets its index, with 60% on private tests or held-out answers
Artificial Analysis’s v5 overhaul changes the benchmark mix and makes 60% of the index depend on private tests or held-out answers.