Every month I do the same research routine for this edition. Most of the time, I already know the headlines. This month I didn't. Two items caught me off guard, and both matter more than the model releases that dominated the AI headlines.

The first is a pricing shift that most finance teams haven't noticed yet. The major enterprise software vendors are quietly moving from per-seat to per-action billing for AI agents. I'm already getting calls from clients who can't explain why their AI bill keeps climbing. That problem is going to get worse.

The second is regulation. The EU AI Act just moved some deadlines, the SEC has a compliance deadline in three days, and AI washing is now personal liability for CFOs. This is too complex to navigate solo. Get legal or compliance counsel involved, build in a review of the regulatory landscape at least every three months, and don't assume a quarterly check-in is enough.

The free section covers all six stories. Paid subscribers get the action items.

But before we move on:

The fourth and last part of the CFO Skills for Claude pack: FP&A and Budgeting.

And if you missed the previous editions, download them here: Financial Reporting, Cash Flow Management, and Audit and Controls.

We are in the middle of the first Claude in Action cohort. We have an incredible group in the room, and I am excited about what the next few weeks look like.

The July cohort waitlist is now open. If you want in at the early pricing, sign up here.

If your company needs something more structured, book a call with me. I run a three-session program as a starting point, but in most cases, we build something specific to your team: the tools you are actually using, the workflows that matter, and the gaps that are costing you time right now.

May looked like a capabilities month on the surface. It was actually a governance month. The stories that will cost you money or create liability are the regulatory and operational ones.

The SEC's amended Regulation S-P compliance deadline for smaller investment advisers (under $1.5 billion AUM) and certain broker-dealers is June 3, 2026. The rule requires a formal incident response program, written safeguarding policies, and individual notification within 30 days of a breach. The SEC's 2026 examination priorities also single out AI: whether firms can explain AI-driven decisions, and whether public AI capability claims are accurate.

Read more: Sidley Data Matters, April 30, 2026

So what: Hard deadline, not a soft guideline. Confirm your AI vendors and data flows are covered, document your tool selection rationale, and get your incident response documentation current before Friday.

On May 7, the EU Council and Parliament agreed to postpone high-risk AI system compliance deadlines. Use-based high-risk systems (credit scoring, employment tools, financial services) move to December 2, 2027. Product-embedded systems get until August 2028. The original August 2026 deadline technically holds until the agreement is published in the Official Journal, expected by July. Draft classification guidelines are out for consultation until June 23.

Read more: Council of the EU, May 7, 2026

So what: This is a planning window, not a pass. If your finance or HR AI tools touch credit decisions or financial assessments, classification applies to you. Use the window now, before the guidelines are finalized and your options narrow.

OpenAI made GPT-5.5 Instant its default on May 5, citing a 52% reduction in hallucinations on high-stakes prompts. Google launched Gemini 3.5 Flash and an agentic Workspace assistant at Google I/O on May 19. Anthropic released Claude Opus 4.8 on May 28, 42 days after Opus 4.7, same price point, with improvements to reliability and honesty rather than capability. All three are building toward agentic execution: AI that plans and completes multi-step work.

Read more: OpenAI, May 5; Google I/O, May 19; Anthropic, May 28, 2026

So what: Vendor-reported hallucination reductions are self-assessed, not independently audited. "Less likely" is not "never." Human review of AI outputs in close, reporting, and disclosure remains mandatory. More practically: the default model changes every six weeks. Build your workflows around tasks, not tool versions.

ServiceNow, SAP, and Workday all moved in May toward consumption-based billing for AI agents. ServiceNow's new Action Fabric charges metered currency per action an external AI agent performs. SAP updated its API policy to bar unsanctioned third-party agents. Workday signaled similar moves. NetSuite's 2026.1 release embedded AI in close management, reconciliations, and bank-transaction matching.

Read more: ServiceNow, May 5; PYMNTS, May 2026

So what: One agent can generate thousands of API calls without adding a single seat. If your AI budget is built on per-seat logic, it won't capture this. The May 27 edition of this newsletter covered the AI cost management framework in detail. The specific gap that edition didn't anticipate is agent-based billing inside ERP and ITSM platforms.

In April and May 2026, two securities cases moved AI claims to the center. A shareholder class action in the Northern District of California alleged a fintech lender misled investors about its AI approval model and raised revenue guidance twice on those claims before missing results. Federal prosecutors also unsealed a 10-count indictment against the former CEO and CFO of iLearningEngines for using fabricated AI contracts to inflate revenue. The SEC, DOJ, and FTC are applying standard fraud frameworks, not novel legal theories.

Read more: Troutman Pepper Locke, May 26, 2026

So what: Every AI capability claim in your earnings call, MD&A, or investor deck is potential securities exposure. And for CFOs, it's personal.

Microsoft's 2026 Work Trend Index (20,000 users, 10 countries, Microsoft 365 telemetry) found that organizational factors account for 67% of reported AI impact. Only 26% of AI users say their leadership is clearly aligned on AI strategy. Only 13% are rewarded for reinventing work with AI when results aren't immediate. Microsoft does not tie any of this to financial payback. That absence is itself a finding.

Read more: Fortune / Microsoft Work Trend Index, May 11, 2026

So what: ROI comes from process redesign and leadership alignment, not license counts. Most companies aren't measuring financial impact from AI at scale. Neither are the vendors selling it. That gap is worth naming in your next board conversation.

We covered the news and the immediate "so what."

The paid section is the action layer: three groups of specific steps with templates and language you can use this week.

May was a lot. Three model releases, a meaningful regulatory shift, securities enforcement making AI claims a personal CFO liability, and a pricing model change most AI budgets aren't built for.

The through-line is the same one that's held for two years. The teams getting real returns from AI are the ones that govern it: a working group, a budget process, a review cadence, and someone who checks the claims before they go out.

If you're working through Reg S-P readiness or starting to structure your AI working group, I'd like to hear what you're seeing. Reply here or find me on LinkedIn.

Until next Tuesday.

Anna

This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!

Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation!

Did you find this newsletter helpful? Forward it to a colleague who might benefit!

Until next Tuesday, keep balancing!

Anna Tiomina
AI-Powered CFO