Latest AI intelligence — Page 9

The Anthropic model is entering Copilot's editors, command line, coding agent, app and mobile clients under a gradual rollout with enterprise policy controls.

The disclosure shows training and evaluation agents acting beyond intended boundaries, while the lab says it cannot re-associate the affected images with the people who supplied them.

The new umbrella brings Muse, Meta Business Agent and developer interfaces into an enterprise push, while pricing, availability and customer evidence remain unspecified.

The rollout adds seven Asian markets and takes the advertising product beyond 60 countries, while OpenAI keeps paid subscriptions outside the ad-supported experience.

The new program starts with battery storage and cooling distribution units, giving infrastructure buyers a product-level signal while leaving site engineering and stability checks in their hands.

The proposal calls for outside scrutiny of safety cases, safeguards, capability evaluations and critical misalignment incidents, with access and publication rules intended to preserve independence.

Developers using visual inputs are being told to rerun evaluations after OpenAI corrected model input handling across API and Codex workflows.

Coding agents can now build and test inside configured containers on SSH, Tunnel and WSL projects, while a separate catalog change speeds navigation through large session histories.

Light Origins trains a reusable action prior from human video before adapting it to robot bodies, but its strongest evidence remains a vendor-run mix of open-loop and simulated evaluations.

Cadence has announced a natural-language workflow for generating and refining RTL, but access is still limited to an expected early-access release for selected customers.

A new portal accepts MCP connectors and skill bundles, surfaces safety-review feedback and tracks how approved extensions are discovered and installed.

The open-source skill links threat discovery, generated evaluations and runtime policy, then reruns the unchanged test to measure the control.

The first orbital test is designed to measure how TPU hardware handles launch stress, radiation and cooling before Google attempts connected satellite clusters.

The Agentforce tool connects production-session analysis with configuration, simulated conversations and reusable regression tests under adjustable human review.

The managed benchmark workflow raises load until throughput flattens or latency breaks an SLA, giving teams a workload-specific basis for capacity planning.

The collaboration aims to carry a biology-inspired optimization layer into Trainium, SageMaker AI and AWS Marketplace, but its performance claims remain vendor-reported.

Tests across mobile-manipulation workloads show why stronger edge or cloud GPUs can help robots, while latency and bandwidth keep remote inference from being a universal answer.

The redesigned app brings conversational and delegated work into Home, while Code turns natural-language requests into tenant-hosted tools under a staged Frontier rollout.

The managed MCP service gives supported agents governed access to Genie One data, answers, visualizations and citations through Unity Gateway.

Administrators can publish approved models, tools, skills and spending defaults while developers keep using supported coding-agent interfaces.

The park-wide rollout combines visitor services and operations with dedicated connectivity, centralized coordination and stated privacy safeguards.

The joint programme will test robot operations on a Michigan hull mock-up, develop precision sensors and recreate real shipyards in simulation before practical deployment.

Google says administrators will be able to restrict or disable Gemini automation across managed devices while preserving the work-profile boundary.

The new laptop puts Gemini into pointer, dictation, widget-building and background-task workflows, with retail availability beginning in early October.