AIが期間内の動向を整理
From Using AI to Running Work Over Time: Latest Moves from OpenAI, Anthropic, and Google Cloud
June 15 highlighted a shift toward treating AI as a system that can keep working over time, not just as a one-off response tool. OpenAI outlined a direction for Codex to support longer-running work, Anthropic announced efforts that combine training and deployment, and Google Cloud emphasized infrastructure for sensitive AI workloads.
SOURCE CHECK
Primary Sources 4
Primary Sources
Key Points
- 1OpenAI outlined a plan to bring Ona’s persistent environments into Codex for longer-running work
- 2Anthropic announced Claude Corps, a plan to place 1,000 fellows at nonprofits
- 3Anthropic partnered with DXC Technology to support Claude adoption in regulated industries and train FDEs
- 4Google Cloud described the infrastructure behind Apple’s Private Cloud Compute and highlighted confidential computing
OpenAI is making Codex fit continuous work
OpenAI announced plans to acquire Ona and bring secure persistent environments into Codex. The point is to let work continue across hours or days rather than ending in a single exchange. In development and routine automation, the ability to preserve context is often what determines practical usefulness.
Anthropic is combining training and deployment
Claude Corps is a plan to place 1,000 fellows at nonprofits. It is designed not just to teach AI use, but to connect it to real work in the field. In the DXC Technology partnership, Anthropic also pairs deployment support for regulated industries with FDE training, showing a strong focus on implementation.
Google Cloud is supporting sensitive AI processing
Google Cloud said it built the serving platform for Apple’s Private Cloud Compute on Google Cloud. The update highlights confidential computing, with protection at rest, in transit, and during use. For companies handling sensitive data, infrastructure choices matter as much as model selection.