Hosts: James & Maya
In this episode:
• Hi, I'm Maya, and today we start with the most expensive AI election yet — and the candidate who won it, then turned on his funders.
• Right. Micah Lasher took New York's 12th district. Anthropic and OpenAI spent a
Pivot5 | 5 Headlines & Unprompted
James: Hi, I'm James.
Maya: Hi, I'm Maya, and today we start with the most expensive AI election yet — and the candidate who won it, then turned on his funders.
James: Right. Micah Lasher took New York's 12th district. Anthropic and OpenAI spent a combined $27 million through super PACs to shape that race.
Maya: And in his victory speech, Lasher rejected both companies and pledged to regulate them. So the money bought attention, not loyalty.
James: Here's why this changes everything for AI companies wading into politics. Direct election spending can now backfire in public.
Maya: Worth noting the caveats here. It's one race. But the lesson is clear — candidates who take tech money may need to publicly distance themselves to look independent.
James: Which makes that $27 million look less like influence and more like a liability. Business leaders watching this should expect AI firms to rethink how they engage politically.
Maya: Onto the markets. OpenAI is reportedly considering delaying its IPO to 2027, and that news moved enterprise software stocks.
James: Salesforce and ServiceNow shares rose. The thinking is simple — if OpenAI isn't going public soon, the competitive threat to incumbents feels further away.
Maya: The data tells a different story underneath the headline. The delay reflects investor skepticism about AI valuations broadly, not just OpenAI's timeline.
James: So incumbents get a breather, but it's tied to sentiment that could shift fast. If confidence in AI valuations returns, that pressure comes back.
Maya: Exactly. Public market mood now connects OpenAI's calendar directly to how Salesforce and ServiceNow are priced. That link cuts both ways.
James: Now a story every CFO should hear. An Emburse survey found four in ten U.S. employees have used AI to generate a fake expense receipt.
Maya: And AI-generated fakes now make up 71% of all expense fraud. Among those who fabricated, 19% invented purchases that never happened, 15% inflated real ones.
James: The real story isn't the fraud itself. It's that traditional receipt checks can't catch machine-made receipts. They look legitimate.
Maya: Let's look at what actually happened. The fraud moved from human-readable to machine-generated. So platforms like Emburse now have to build AI detection into their core controls.
James: If you run a finance team, assume your current verification is already behind. This is a near-term problem, not a future one.
Maya: Our fourth story is a quieter one. A researcher used Claude Opus to create a self-portrait, then laser-engraved it into a physical object.
James: What nobody's talking about yet is that model identity is becoming a design material. Claude's persona was the creative input, not just the chat.
Maya: It's early and experimental. But it hints at artist and research uses for AI personas beyond the usual products — and maybe commercial ones later.
James: Last in the main segment, a warning from QBE North America and Control Risks. Agentic AI is speeding up commercial espionage and fraud.
Maya: Their report says financial institutions racing to deploy autonomous agents face a widening gap between deployment speed and threat modeling.
James: The same agents that promise efficiency create new ways in for attackers, moving at machine speed. Expect insurers to tighten underwriting around these systems.
Maya: Worth tracking how quickly risk frameworks adapt. Right now, deployment is outrunning the security thinking behind it.
James: That's our look at today's stories — the election money that backfired, the IPO delay, AI expense fraud, model identity as art, and agentic AI security risks.
Maya: Five threads, one common point — AI is now reshaping politics, markets, finance controls, culture, and security all at once.
James: Plenty to act on this week.
Maya: That's your Pivot briefing. We'll be back with more analysis soon.