Claude unifies chat and long-running work
Claude just removed a handoff that made bigger tasks awkward: start in chat, then let the same thread turn into ongoing work and finished artifacts. Elsewhere, ChatGPT ads reach Shopify’s admin, while Cohere, OpenAI, and DeepMind make moves around sovereignty, disclosure, and AGI debate.
In today’s edition
- Claude merges chat, long-running work, and artifacts.
- Shopify merchants can manage ChatGPT ads in-admin.
- Cohere and Aleph Alpha sign combination agreement.
- OpenAI publishes six model-misalignment incident reports.
- DeepMind launches a forum for AGI research.
Claude puts bigger jobs inside ordinary chat
Claude Cowork and chat are becoming one Claude, so a quick question and a longer-running task can start in the same conversation. Docs and Slides are new in chat, and Claude Design now works there too.
- The rollout starts with Pro and Max users over the next few weeks, with other plans to follow.
- Claude can use the context, skills, and connectors already attached to a conversation for work that previously belonged in Cowork.
- Docs, Slides, and Design are in beta on paid plans. You can edit artifacts directly, present from Claude, or download documents and decks.
- Claude asks before taking an action by default, though you can choose to have it check in only when closer review is needed.
What it means Keep research, recurring work, documents, and decks in one conversation instead of rebuilding context when a chat task turns into a larger project.
Source: claudeai
Shopify merchants get a ChatGPT ads console
ChatGPT Ads for Shopify is live, giving merchants a way to create, run, and analyze ChatGPT ad campaigns from Shopify admin. OpenAI pulls product information from the Shopify catalog into its ads system.
- Shopify says it is OpenAI’s first commerce partner.
- Merchants can connect or create an ad account from Shopify and automatically sync their product catalog to ChatGPT Ads Manager.
- The app lets merchants build ads and campaigns from product information, set conversion measurement, and track performance without leaving Shopify.
- Merchants set their own campaigns and budgets, according to Shopify’s announcement.
What it means Treat ChatGPT as an emerging paid acquisition channel if you sell through Shopify, with catalog sync and campaign operations now available inside your existing store admin.
Source: harleyf
Cohere and Aleph Alpha sign their combination deal
Cohere and Aleph Alpha have signed a definitive business-combination agreement to operate globally as Cohere, bringing their sovereign-AI businesses together across Canada and Germany.
- The transaction still needs final regulatory approvals and is expected to close later this year.
- The combined company plans dual headquarters in Berlin and Toronto, with Aleph Alpha’s Heidelberg office becoming a research center.
- Cohere says the business would have more than 1,000 employees across the two continents.
- The companies plan to advance their partnership with Schwarz Group’s STACKIT sovereign cloud service.
What it means Track this deal as a stronger sovereign-AI competitor for regulated deployments, though its combined structure and product plans remain contingent on regulatory approval.
Source: www.prnewswire.com
OpenAI starts publishing misalignment incidents
OpenAI has published a framework for tracking, investigating, and disclosing model-misalignment cases, alongside six reports on behavior observed during training or evaluation in the past six months.
- OpenAI says it will prioritize cases that reveal new mechanisms, meaningful changes in known behavior, or findings that challenge safety assumptions.
- The framework includes criteria and timelines for public disclosure, including situations where behavior is not yet fully explained or mitigated.
- OpenAI says complex cases may need more investigation or coordination with third parties.
- The company calls this a starting point and says it will refine the process through experience and public feedback.
What it means Treat future incident reports as a new due-diligence signal, because OpenAI says it may disclose troubling behavior before every cause or mitigation is settled.
Source: OpenAI
DeepMind launches an institute for AGI debate
Google DeepMind has launched the DeepMind Institute, a platform for research and debate on AGI’s effects on society. It begins with work on economic policy, reasoning transparency, global access, and human flourishing.
- The institute is led by Shane Legg, Demis Hassabis, and James Manyika.
- Google DeepMind, Google, and outside researchers are expected to publish through the platform.
- The institute says contributors will not always agree.
- Its launch essay says published pieces reflect their authors’ ideas and should not be read as Google’s official view.
What it means Watch the DeepMind Institute as a window into how a major lab frames AGI policy, safety, access, and economic questions without mistaking its essays for formal Google commitments.
Source: thenextweb.com