EU eyes AI Act to police frontier labs as Anthropic merges its agent tools
The EU's von der Leyen wants global safety talks with AI labs, Anthropic combined its chat and agent products, and Apple is reportedly building an AI server for 2029.
Every line links to a primary source in the full briefing.
01EU wants a seat at the table on global AI safety rules
European Commission president Ursula von der Leyen said she plans to invite major AI labs to safety talks and push the EU's AI Act as a model for international standards. She pointed to risks like AI agents acting outside their intended boundaries and self-improving models as reasons for urgency.
If the AI Act becomes the reference point for global rules, compliance requirements that already apply to EU businesses using AI tools could become the default expectation from vendors everywhere.
Source: The Decoder
02Anthropic combines Claude chat and its Cowork agent tool
Anthropic merged its standard Claude chat interface with Cowork, its agent-style workspace tool, into a single product. The combined interface is rolling out first to Pro and Max plan subscribers.
Businesses already paying for Claude get agent-style task automation folded into the tool they use daily, lowering the barrier to trying it without a separate subscription or setup.
Source: TechCrunch
03OpenAI builds advertising tools directly into ChatGPT
OpenAI unveiled new AI-powered advertising features including something it calls Sponsored Agents, plus tools aimed at marketers and integrations with HubSpot and Shopify.
Small businesses running ads or e-commerce through Shopify or HubSpot may soon see AI-driven ad placement and targeting options appear inside tools they already use, worth watching before adopting.
Source: OpenAI
04Apple reportedly designing its own AI server hardware
Apple is reportedly building a server packed with its M-series Ultra chips specifically for AI workloads, with a planned debut around 2029. This would reportedly be Apple's first enterprise server product in decades.
This is a long-horizon hardware bet, not something that changes any business's options today, but it signals Apple wants a bigger role in AI infrastructure rather than just devices.
Source: Ars Technica
05Mozilla brings a Mistral-powered assistant into Firefox
Mozilla is testing a new 'Smart Window' assistant feature in Firefox, running on Mistral's Small 4 model, currently in beta. It marks Mozilla's most direct move to build AI assistance into the browser itself.
Businesses that rely on Firefox may soon get built-in AI browsing help without needing a separate extension or subscription, though it is still early-stage and in beta.
Source: the-decoder.comPulse 2.0
06OpenAI pitches tools to measure AI's return on investment
OpenAI published guidance on using ChatGPT Work and Codex analytics to track how teams use AI, spot where more training is needed, and tie usage back to business outcomes.
For businesses that have rolled out AI tools but struggle to justify the spend, this points to built-in analytics rather than needing a separate measurement system.
Source: OpenAI
Sources
7 links- 01EU president warns AI agents "escaping their environment" are just a preview of what's comingThe Decoder
- 02Reimagining advertising with AIOpenAI
- 03How to connect AI usage to business valueOpenAI
- 04Apple reportedly building server packed with M-series Ultra chips for AIArs Technica
- 05Anthropic merges Claude chat and Cowork in one interfaceTechCrunch
- 06Mozilla's new Smart Window assistant runs on Mistral's modelsthe-decoder.com
- 07Mozilla Partners With Mistral To Bring Mistral Small 4 To Firefox Smart Window BetaPulse 2.0