
- 1. Microsoft launches Copilot mode for agentic browsing
- 2. Google adds new features to AI mode
- 3. China pitches its own AI action plan
- 4. OpenAI backs legal rights for AI, and the right to be forgotten
- 5. Tesla signs $16.5B AI chip deal with Samsung
- 6. Alibaba unveils AI smart glasses
- 7. NotebookLM adds AI-generated video summaries and Studio upgrades
- 8. Anthropic limits Claude code use for power users
- 9. Zuckerberg says superintelligence is “now in sight”
- 10. Palo Alto Networks to acquire CyberArk for $25B
- Recommended reading:
The CFO AI Digest: July 31
This week: Google and Microsoft push AI deeper into everyday workflows, Tesla doubles down on custom chips, and China enters the global AI rulemaking arena. Meanwhile, new limits, new devices, and new apps are quietly shifting how and where AI creates value.
1. Microsoft launches Copilot mode for agentic browsing
Copilot mode turns Microsoft Edge into an AI assistant that goes beyond search. It actively interprets browsing behavior, compares across tabs, summarizes articles, and can take actions for the user, like filling forms or navigating pages, based on context.
CFO takeaway: Agentic browsing is coming to finance software whether teams are ready or not. Tools will act on behalf of users online—sometimes well, sometimes not. Think: form fills, password resets, even misfired CAPTCHAs. Teams need guardrails in place, and policies that assume “bot” activity might just be your own people using AI.
2. Google adds new features to AI mode
Google is expanding its AI-powered “Overviews” in Search to deliver direct, conversational answers instead of just showing websites. New features include improved summarization, suggested follow-up questions, and the ability to adjust tone.
CFO takeaway: AI-driven search will cut traffic to traditional websites and disrupt how companies are found. If your business relies on search for leads, marketing, or recruiting, expect rising acquisition costs and diminishing SEO returns.
3. China pitches its own AI action plan
Days after the U.S. unveiled its national AI strategy, China launched a Global AI Governance Action Plan at the World AI Conference in Shanghai. Premier Li Qiang called for a United Nations–led system to oversee AI development, emphasizing fairness, equity, and respect for national sovereignty. The plan includes a new “AI for Good” initiative and a pledge to support developing countries with access to AI technologies and infrastructure.
CFO takeaway: AI governance is now geopolitical. China’s multilateral, state-first model challenges U.S. and EU norms. Expect more compliance friction, fragmented standards, and mounting pressure to localize data and AI infrastructure.
4. OpenAI backs legal rights for AI, and the right to be forgotten
OpenAI now advocates for user protections like the right to be forgotten when interacting with AI systems. The company also suggests that, in the future, highly autonomous AI agents may require a form of legal status to act on users’ behalf. It stops short of calling for full legal “personhood” but opens the door to limited legal agency for AI.
CFO takeaway: This pushes AI into uncharted legal territory. Corporate personhood lets companies enter contracts and be held accountable. Now imagine a system where AI agents can do that, without clear lines of intent, delegation, or liability. That shift would reshape compliance and audit standards. Finance teams will need to track not just what the AI did, but why, and who authorized it.
5. Tesla signs $16.5B AI chip deal with Samsung
Tesla is partnering with Samsung to produce next-gen AI chips for Full Self-Driving, robotics, and Dojo (Tesla’s in-house supercomputer designed to train AI models for autonomous driving). The chips will use Samsung’s advanced 4-nanometer manufacturing process, which allows for faster performance and lower power use. Production will roll out over five years, making this one of the largest and longest-running chip deals in the industry.
CFO takeaway: Last week's Digest was all about this kind of vertical integration: big name companies are trying to get a foothold by bringing AI stacks in-house. Expect to see more of this in the weeks to come, followed by tightening chip supply, rising costs, and shifting control over AI interfaces.
6. Alibaba unveils AI smart glasses
Alibaba’s AI-powered smart glasses are equipped with a built-in camera, real-time translation, and computer vision to recognize objects and text. They’re aimed at visually impaired users, but signal Alibaba’s broader play into AI hardware, following Meta, Apple, and ByteDance.
CFO takeaway: AI hardware is becoming a strategic priority. Alibaba’s entry signals that advanced optics, on-device processing, and language AI are becoming core battlegrounds. Platforms outside the U.S. are now shaping where user data and spending go next.
7. NotebookLM adds AI-generated video summaries and Studio upgrades
NotebookLM, the AI-powered research assistant by Google, now offers video summaries, letting users upload videos and get AI-generated overviews. Separately, “NotebookLM Studios” create custom environments where teams can turn their own content into AI-powered workspaces.
CFO takeaway: Google is turning AI summarization into a product, as NotebookLM evolves from note-taking to enterprise workflows. This opens a new use case: organizing and extracting value from internal data and content.
8. Anthropic limits Claude code use for power users
Anthropic is introducing new rate limits for Claude, aimed at curbing excessive use of its coding capabilities. The company says a small group of users has been consuming a disproportionate share of compute resources by automating code generation at scale.
CFO takeaway: Foundation models aren’t free to scale. Anthropic’s limits highlight the real cost of inference, especially for high-volume tasks like code. Power users may face higher pricing or tighter restrictions as providers balance cost, performance, and profitability.
9. Zuckerberg says superintelligence is “now in sight”
In the Meta earnings call, CEO Mark Zuckerberg said that achieving artificial general intelligence (AGI) or “superintelligence” is now within reach. He positioned it as a core goal, with open-source models like Llama accelerating progress.
CFO takeaway: Over the past six months, leaders at Anthropic, OpenAI, and Meta have framed AGI not just as inevitable, but imminent. Regardless of whether that timeline proves accurate, these declarations are raising the stakes. They’re fueling urgency, driving boardroom pressure to stay competitive, and amplifying the race to adopt. Yet for most teams, the real advantage isn’t in what might be possible tomorrow—it’s in what can be reliably deployed today.
10. Palo Alto Networks to acquire CyberArk for $25B
Palo Alto Networks is acquiring Israeli cybersecurity firm CyberArk in a $25 billion deal, its largest ever. This is a direct response to rising AI-driven cyber threats and a shift toward identity-centric security. CyberArk specializes in privileged access management, a key control point as AI agents begin executing tasks across systems.
CFO takeaway: As AI expands inside organizations, identity becomes the new perimeter. This deal signals that traditional endpoint security isn’t enough; AI-powered attacks require AI-aware defenses. Finance teams should expect rising spend on identity and access controls, especially as autonomous systems gain privileges once reserved for humans.
See you next week.