The Briefing Table
May 2026
Welcome to the May issue of The Briefing Table where we review each month’s major events from a strategic and risk management perspective.
April was a month where a familiar pattern reasserted itself. A ceasefire that was supposed to restore normal conditions didn’t. A tariff deadline that was supposed to be about pharmaceutical companies turned out to affect almost every large employer in the country. An AI model arrived that its own developer deemed too dangerous to release. A regulatory deadline that had been on the calendar for two years became a deal-room variable. And a Federal Reserve succession that was supposed to be a procedural formality became the first fully partisan confirmation vote for a Fed chair in American history.
These five forces, each a deferred cost arriving on a compressed timeline, are the subject of this issue.
1) Anthropic Mythos: The Threat That Wasn’t Supposed to Arrive This Year
Ask most CIOs what they expected from AI and cybersecurity in 2026, and the answer would have been incremental. A genuine capability breakthrough, serious enough to prompt a major AI lab to withhold a model from the public entirely, was not in the forecast.
It arrived anyway.
On April 7th, Anthropic announced Claude Mythos Preview and declined to release it publicly. Whereas OpenAI’s temporary suppression of GPT-2 in 2019 was widely derided as theatrical, no one is mocking Anthropic. During testing, Mythos identified and exploited zero-day vulnerabilities across every major operating system and browser, including bugs that had survived decades of human review. The UK’s AI Security Institute found it succeeded at expert-level hacking tasks 73% of the time. One year ago, no AI model could complete those tasks at all.
In response, Anthropic launched Project Glasswing, committing $100M in usage credits to let AWS, Cisco, CrowdStrike, and forty other organizations use Mythos defensively to scan critical infrastructure before adversaries could exploit the same capabilities. Anthropic did not train Mythos to have offensive security capabilities. They emerged from general improvements in reasoning and code, which means every subsequent frontier model will have them whether its developer intends it to or not. The controlled release window is also shorter than it appears: independent researchers have shown open-weight models just three to five months behind the frontier can replicate much of the same analysis. U.S. regulators have drawn their own conclusion and are already urging major financial institutions to begin controlled testing now.
The temptation is to file this under “cybersecurity” and hand it to the CISO. That would be a mistake. Mythos-class discovery will flood enterprise vulnerability queues with findings that security teams have no additional capacity to remediate. More acutely, the agentic systems companies are deploying to automate workflows are now the most valuable targets for adversaries. A single prompt injection can co-opt an organization’s highest-privileged digital actor. Last month’s issue argued that the risk in agentic AI is in what agents do, not what they are. Mythos makes that argument viscerally. Ungoverned agents are unlocked doors, and the lockpicks just got dramatically better.
There is also a competitive dimension many C-Suite executives are missing entirely. The organizations currently using Mythos-class capabilities defensively are building a security posture their peers will not have for another year or two. This is the same structural dynamic that played out with every previous generation of security infrastructure. The firms that built the capability before the breaches gained lasting advantages. The ones that waited explained the gap to senior executives. The window to be the former rather than the latter is open now. It will not remain so.
Bottom Line: Commission an agentic AI inventory audit now; what agents are running, what systems they can access, what actions they can take without human review. The organizations that act in the next ninety days will be the ones that don’t have to explain why they didn’t.
2) Pharmaceutical Tariffs: This Is Not a Healthcare Story
Most coverage of the April 2nd pharmaceutical tariff proclamation has treated it as a healthcare sector story. It is not. It is an employer story, a supply chain story, and a strategic optionality story with a deadline.
Under Section 232, the administration imposed 100% tariffs on imported patented pharmaceuticals and APIs, effective July 31st for large companies. The structure rewards speed: a 0% rate through 2029 for a MFN pricing agreement with HHS and a domestic onshoring plan with Commerce versus 20% for an onshoring plan only. Do neither and face 100%. This is the sectoral escalation we previewed in February’s edition under “The End of Emergency Tariffs.” Parallel Section 232 investigations are already underway in medical consumables, devices, and robotics – i.e., the pharmaceutical action is a template, not a one-off.
Importantly, the MFN/onshoring path is not a compliance exercise. It is a negotiation, and the tariff framework is effectively offering a four-year subsidy, zero tariff liability, in exchange for policy alignment. Organizations recognizing this and moving first will secure favorable terms. Those treating it as a burden to manage will pay the default rate. In addition, the majority of large U.S. companies self-insure their health benefits. Specialty biologics and patented therapies facing the 100% default represent a disproportionate share of those plans’ costs. This is a human capital cost problem, not a healthcare problem, and it arrives before year-end.
Bottom Line: Three parallel tracks: classify your exposure across patented pharmaceuticals and APIs; assess the MFN/onshoring compliance path, because the zero-rate tier requires simultaneous engagement with two federal agencies before July 31; and complete the supply chain audit begun in April; Chinese and Indian API production is a dependency as significant as Gulf-region helium and sulfur.
3) The Iran Ceasefire: Beijing Is the Story Now
Last month, we wrote: “Do not treat a sudden ceasefire as a return to normal.” A ceasefire arrived within two weeks and has been violated by both parties. As of publication, Iran submitted a formal proposal to reopen the Strait in exchange for the U.S. lifting its naval blockade, while deferring nuclear discussions, but Washington effectively rejected it. Meanwhile, Pakistan has stood down the security infrastructure it assembled for a second round of talks and there is no next meeting scheduled. The new baseline is not a ceasefire but a siege.
Just below the surface, a new regional power structure is taking shape. China did not merely benefit from this conflict. It is now shaping its resolution. Beijing’s co-mediator role in Islamabad was confirmed by the White House. Iran’s oil exports have been settled increasingly in renminbi throughout, a practical arrangement not reverting when hostilities end. The country helping end a war earns diplomatic capital persisting for years. China is accumulating it while the United States absorbs the costs.
Jeff Sachs observed in his April interview with C-Suite that the dollar’s share of global reserve holdings has been declining for two decades and that the pace is accelerating. The Iran conflict has now added a live demonstration that the renminbi can function as a credible settlement currency in the world’s most strategically sensitive energy corridor. Sachs’s projection - renminbi exceeding 20% of international trade settlement within a decade, the dollar falling below 50% - now feels less like a long-range forecast and more like an extrapolation of something already in progress. The right move for corporate treasurers, he advised, is to begin building renminbi-denominated banking capacity now. China’s central role in the ceasefire process makes the timeline feel urgent in a way that abstract reserve currency projections do not.
The practical prescription is the one from last month, now sharper: build explicitly for both endpoints. A rapid normalization would compress oil prices, normalize shipping, and release Gulf sovereign wealth into global markets. The 14% intraday swing in early April showed how fast transitions happen. The side you are not prepared for is the one that will cost you.
Bottom Line: The Iran story has two chapters. The first has been modeled: Hormuz disruption, energy costs, supply chain exposure. The second typically has not: China as indispensable regional broker, renminbi advancing in energy settlement, dollar primacy eroding in the one market where it was most entrenched. It should be.
4) The EU AI Act: Already a Valuation Variable
The EU AI Act’s August 2nd enforcement deadline is now approximately 100 days out, and European deal rooms are already pricing compliance posture into transactions. One Q1 transaction closed at a €7 million discount written into the SPA as a specific AI Act indemnity. One HR analytics asset was pulled from market after four bidders flagged the same Annex III gap. An Austrian lender commanded a two-turn valuation premium because its credit model had been documented against Article 10 since late 2024.
High-risk categories, including hiring, credit, biometric identification, and critical infrastructure, require conformity assessments, technical documentation, CE marking, and EU database registration before August 2nd, with penalties reaching €15 million or 3% of global turnover. The Act applies extraterritorially.
Over half of organizations still lack systematic AI inventories. Without an inventory, risk classification is impossible. Conformity assessment takes six to twelve months. The arithmetic is unambiguous for anyone who has not started.
For organizations conducting M&A or operating in EU-regulated sectors, the compliance posture is no longer a legal question. It is a valuation question. And it lands on the same workforce managing every other demand in this issue. This publication’s March interview with Sponge Group found roughly 60% of employees describe themselves as ‘okay’ or ‘indifferent’ - what they called ‘Generation Numb’ - with change capacity quietly depleted across the organization. The executives who will deliver AI Act readiness within 100 days are not those who issue another mandate from the top. They are those who give their teams agency in how it gets done.
Bottom Line: Start with the inventory. For M&A teams, the question to every diligence counterparty is: show me your AI registry and risk classification. The difference between a two-turn premium and a seven-figure indemnity is documentation that either exists or doesn’t.
The Fed: A Convention Has Ended
Every Fed chair confirmation has included bipartisan support. That norm is gone now, with Kevin Warsh clearing the Senate Banking Committee on April 29th on a straight party-line vote, enabled by the DOJ’s decision to drop its criminal investigation into incumbent Chairman Powell to clear Senator Tillis’s hold. That sequence set a precedent: threatening a sitting Fed chair with criminal prosecution is now an established executive branch tool. The Fed held rates steady at its final Powell meeting. Futures markets price a largely steady path through 2027. Trump wants 1%. That gap is the uncertainty premium embedded in every capital allocation decision right now.
The Federal Reserve Act is deliberately ambiguous on presidential removal authority. It has rested on political convention: an implicit agreement between both parties that the economic cost of undermining the central bank’s credibility exceeded any short-term political gain. That agreement ended on April 29th. Whether Warsh honors the substance of that convention while the form of it has been broken is now the central question for anyone managing dollar exposure, long-duration assets, or capital allocation decisions that depend on reliable inflation anchoring over the next decade.
Bottom Line: Watch Warsh’s first public statement. Does he frame the Fed’s mission in terms of the president’s agenda or its dual mandate? The former signals a repricing of dollar assets and U.S. sovereign risk. The latter suggests the guardrails held. Either way, five decades of institutional convention ended April 29th.
Key Take-Aways
Mythos has crossed a capability threshold the industry didn’t expect this year. The defensive window is months, not years. Ungoverned agents are unlocked doors. Commission the inventory audit.
Pharmaceutical tariffs are not a healthcare story but a self-insured employer story with a July 31st deadline. The compliance path is a negotiation with a hidden reward for speed. Start both tracks now.
The Iran stalemate has hardened into a siege. China is the structural winner. Build for both endpoints as rapid normalization will wrong-foot you as badly as escalation if you’re not prepared.
The EU AI Act has moved from legal to financial. Deal rooms are already pricing the gap. Start with the inventory. Everything else follows.
The Fed succession is not a political story. It is a capital markets story. Watch Warsh’s first statement. A five-decade convention ended last week regardless of what he says next.
All five forces share a structure: deferred costs, compressed timelines, stretched execution capacity. The executives who navigate this environment best will be those who see that pattern and build resilience before the next deadline, not after.