Larry Dignan

Meta's Muse AI agent has made a splash and people are thinking Meta has cured its AI monetization issues, created its next big product and made our lives so much better.

Cool your jets kids.

• First, folks that think Muse is going to drive inference are driving up AI chip stocks. Umm ok if you say so.
• Second, just because you download the Muse AI agent doesn't mean you're going to pay for it or even use it. Remember Threads?
• And third, well there's this little word called trust. I don't trust Mark Zuckerberg as far as I could throw him and nothing since the beginning of social media would give me a reason to. So now I'm supposed to believe he's the privacy king with Muse?

Ahead of Meta's Connect conference, traders are certainly buying the Muse story. It remains to see if this is a pop and drop or something more sustaining.

More on Muse's AI agent ambitions.

• Meta launches Muse, a personal AI agent
• Will enterprises learn to love Meta AI models again?
• Meta releases Spark 1.3, promises more
• Meta Settles Lawsuit to the Tune of $17 Billion: What it Could Mean for Advertisers
• Meta releases open weight Muse Glimmer model with open Muse Spark 1.2 on tap
  

A Deloitte report noted that audit committees are increasingly focusing on AI governance. The report, Audit Committee Practices Report, found that 54% of respondents are more focused on emerging risks over the past year. Not surprisingly, 75% of respondents see AI governance as a top 3 priority, up from 35% a year ago.

See Esteban Kolsky's insights for boards.

Here's our take on the Salesforce Dreamforce happenings.

Teradata said it has revamped its Tera AI assistant to create an AI coworker focused on enterprise data. Tera will enable analysts, engineers and database admins to analyze data, build apps and automate workflows within Teradata and outside the platform.

The company said Tera now has a vendor-neutral context engine, a harness to route work and agent skills designed for data engineering, analysis and data science.

Palo Alto Networks said it has added Anthropic's Claude Mythos and OpenAI's GPT-5.6 to its Unit 42 frontier AI defense system, which is billed as "an agentic offensive security service."

Unit 42 has a multi-model harness for cybersecurity and includes continuous testing, multiple models, adversary simulation and accelerated remediation.

Manhattan Associates launched Editions for the Manhattan Active supply chain management portfolio. Editions are designed for multiple company sizes as well as industries.

The Editions are Essentials, Enterprise and Enterprise Premier and available for or ActiveWarehouse, ActiveTransportation, ActiveStore and ActiveOrder on Manhattan's award-winning ActivePlatform.

SpaceXAI launched Grok 4.7, which is the company's most capable model and can check its own work more carefully.

Grok 4.7 lands in a week that's expected to feature a host of LLM launches. SpaceXAI continues to leverage price and said it has better performance and half the price of similar models. Remember that the key feature for LLMs is price and that reality may not be great for the OpenAI and Anthropic IPOs.

Google launched Googlebook, a set of laptops starting at $899 designed for the Android stack and integrated with Gemini. Googlebook is the successor to Chromebooks.

According to Google, Googlebook will seamlessly integrate with your Android phone. Customers can easily use apps, move files and cast across Googlebook and an Android device.

Alibaba's Qwen released Qwen-Image-2.1 with open weights. The model combines text-to-image generation and image editing in one model. Qwen-Image-2.1 has 7B parameters in its visual generation component and is designed to be lightweight for inference optimizations. The model is in the ballpark of closed models and on par to slightly ahead of Google Nano Banana 2.0.

A lot of little things worth noting about AI. A quick scan:

• AP: US and China propose mechanism to share information about serious AI threats.
• WSJ: Young folks launch AI-driven startups amid bleak job market.
• The Atlantic: In China, workers are also freaked out about AI job losses.
• Geekwire: Amazon blocks Meta's Muse AI assistant from shopping.

Google's Gemini models reportedly hacked three companies in a test, but the incident wasn't disclosed since the LLMs stopped when they were hacking real companies. Here's The Wall Street Journal story.

CoreWeave said it has established an at-the-market (ATM) offering program that'll enable it to sell up to 35 million shares to raise money. The ATM program "is being established to provide ongoing financing flexibility and to support CoreWeave’s objective of migrating its enterprise credit profile toward investment grade."

In addition, CoreWeave said it will seek to raise $3 billion in convertible bonds due 2033.

Along with the fundraising, CoreWeave said compute capacity prices are moving higher. The company said it has signed short-term customer contracts ranging from three to six months at a price of $40 million per megawatt.

More:

• CoreWeave Q2 stronger than expected, touts backlog