AI is creating new opportunities, but also raising difficult questions. Howard Marks is questioning whether investors can profit from AI's growth. Meanwhile, KPMG is expanding Microsoft Copilot to over 276,000 professionals globally. And new research reveals how attackers are impersonating AI agents to bypass traditional defenses. Together, these stories highlight three challenges shaping AI's future: identifying where value will emerge, turning adoption into business results, and securing systems as AI takes on more responsibility.

In today’s post:

• KPMG's 276,000-person AI bet

• AI's biggest risk isn't the technology

• The AI agent you trust might be a bot

SPONSORED BY

How AI-Era Pricing Is Reshaping Finance Operations

Usage-based and hybrid pricing models are changing how B2B companies generate revenue — and creating new headaches for the finance teams behind them.

Tabs co-founder Rebecca Schwartz and PwC Partner Amit Dhir sat down to unpack exactly what that means in practice: how pricing model decisions ripple into revenue recognition, forecasting, and financial ops — and what it takes to scale without piling on manual work.

Watch the on-demand recording to get practical frameworks, real-world examples, and a clear path to operationalizing usage-based revenue — including a forward-looking take on how AI will reshape financial workflows. If your team is navigating pricing complexity heading into the back half of the year, this is worth an hour.

Watch the recording now

What’s Trending Today

GROWTH

KPMG is moving AI from individual tools into everyday work

Microsoft announced an expanded AI collaboration with KPMG. The rollout will cover more than 276,000 professionals globally. But the bigger story isn't the number of users. It's how KPMG plans to manage AI across its business.

Here are seven things worth understanding:

• KPMG is expanding Copilot across its global workforce. More than 276,000 professionals will gain access. The deployment includes KPMG's operations in Thailand.

• Microsoft Agent 365 will help manage AI agents. KPMG plans to monitor, secure, and update agents centrally. This matters as autonomous systems become part of everyday operations.

• The focus is shifting beyond individual productivity. Companies increasingly want AI embedded throughout their workflows. That requires redesigning processes, not simply introducing new software.

• KPMG is using its own deployment to build expertise. Its experience should inform how it advises clients. Real-world adoption exposes challenges that isolated pilots often miss.

• Governance is becoming essential to enterprise AI. Organizations need clear ownership, security controls, and accountability. Scaling agents without oversight introduces new operational risks.

• KPMG Workbench connects multiple AI agents. Built on Microsoft Foundry, the platform supports client service delivery. Its applications span advisory, tax, and audit.

• Business outcomes will determine whether adoption succeeds. Deploying AI doesn't automatically improve productivity or profitability. Companies still need to measure the value created.

There's an important distinction here. Giving employees AI tools is relatively straightforward. Changing how an entire organization works is much harder. The interesting part isn't KPMG deploying Copilot to thousands of employes. It's the decision to build systems around AI agents. Software adoption used to focus on getting people into applications. AI may reverse that relationship. Applications and agents could increasingly perform work on people's behalf. That creates a different management challenge. Companies must decide what AI handles, what humans review, and who's accountable. The next competitive advantage may come from designing those boundaries well.

RESEARCH

Howard Marks asks the question AI investors can't answer

Howard Marks published a new memo on AI investing yesterday. His central concern isn't whether AI will transform the world. It's whether investors will earn enough from that transformation. The distinction matters more than most people realize.

Here are seven ideas worth thinking about:

• A great technology isn't necessarily a great investment. The internet transformed business, but many early investors lost money. AI could follow a similar path.

• Predicting AI's winners is unusually difficult. Investors must correctly anticipate technology, competition, demand, profitability, and timing. Being right about one isn't enough.

• The biggest beneficiaries might be AI's customers. If competing models offer similar capabilities, prices could keep falling. Businesses using AI may capture more value than model developers.

• Data centers aren't automatically safe investments. Electricity shortages, financing costs, outdated chips, and overbuilding create risks. Physical infrastructure doesn't eliminate uncertainty.

• AI has become a concentrated market risk. Marks cites research attributing roughly two-thirds of recent S&P 500 gains to 42 AI-related companies. That creates vulnerability beyond direct AI investments.

• Debt introduces another layer of risk. Marks cites nearly $600 billion in AI-related borrowing this year. Future revenues must eventually justify today's enormous spending commitments.

• The most exciting companies might be the hardest to value. Claude rated frontier AI labs just 2/10 for predictability. Their technological potential doesn't guarantee durable profits.

The deeper lesson isn't that investors should avoid AI. It's that uncertainty should influence what they're willing to pay. Marks makes an important distinction between conviction and certainty. You can believe AI will change everything without knowing who wins. And you can recognize enormous potential without accepting any price. The best investment may not be the most impressive technology. It may be the opportunity where uncertainty is already reflected in valuations. So here's the question worth asking:

AI AGENTS

Attackers are impersonating AI agents, while real agents disguise themselves as humans

A new Cequence Security analysis reveals a growing cybersecurity problem. Attackers are exploiting the identities of legitimate AI agents. Meanwhile, genuine agents increasingly resemble ordinary human visitors. Traditional bot detection struggles to distinguish between them.

Here are seven findings worth understanding:

• One fake AI agent generated 414,210 visits in a week. Every request targeted an online marketplace's login page. The activity was part of a credential stuffing attack.

• The impersonator generated more traffic than genuine agents. Its activity exceeded legitimate agent visits across five industries by 50%. A familiar agent identity offered no guarantee of legitimacy.

• Real AI agents are also hiding behind browser identities. One telecom provider observed automated traffic reaching 19,000–32,000 daily requests. The agent handled logins, bills, authentication, and shopping activities.

• A browser update exposed the automation. Nearly 98% of suspicious traffic adopted the new version within 48 hours. Human customers took roughly two weeks to make that transition.

• Most legitimate AI agent activity remains low-risk. Across 281,627 visits, 85% involved browsing or product comparisons. Roughly one in seven sessions involved more sensitive actions.

• Traditional bot detection missed most unauthorized login attempts. Behavioral analysis blocked 3.51 million attempts at one consumer brand. Around 82% fell below conventional bot detection thresholds.

• Security teams need to monitor behavior, not claimed identity. Repeated login failures should trigger scrutiny regardless of agent branding. Stronger verification should focus on payments and account changes.

The broader implication is becoming clear. Knowing what software claims to be no longer establishes trust. The interesting shift isn't simply that attackers impersonate AI agents. It's that legitimate AI agents increasingly behave like human users. Both can sign in, navigate accounts, and initiate transactions. That creates a problem for security systems built around identity. A browser can be automated. An agent's name can be forged. Neither tells you whether an action is trustworthy. The better question is what the session actually does. As AI agents gain more control, behavioral security becomes essential.

Free Guides

My Free Guides to Download:

🚀 Founders & AI Builders, Listen up!

If you’ve built an AI tool, here’s an opportunity to gain serious visibility.

Nextool AI is a leading tools aggregator that offers:

• 500k+ page views and a rapidly growing audience.

• Exposure to developers, entrepreneurs, and tech enthusiasts actively searching for innovative tools.

• A spot in a curated list of cutting-edge AI tools, trusted by the community.

• Increased traffic, users, and brand recognition for your tool.

Take the next step to grow your tool’s reach and impact.

Reach 150,000+ READERS:

Expand your reach and boost your brand’s visibility!

Partner with Nextool AI to showcase your product or service to 140,000+ engaged subscribers, including entrepreneurs, tech enthusiasts, developers, and industry leaders.

Ready to make an impact? Visit our sponsorship website to explore sponsorship opportunities and learn more!