Latest / Tech IPO Conversations with Fexingo: Public Software Companies and Capital Markets / How Post-IPO Companies Use Insider Trading Bans After AI Incidents
Transcript
- Lucas: Apple filed a lawsuit against OpenAI on Friday, alleging trade secret theft related to its AI models. But what I find interesting is the immediate question it raises for any post ipo company: if you discover an employee used confidential training data against policy, do you have to wait for the next quarterly trading window to claw back their equity or restrict their trades? Luna: Because that delay could be weeks or months. In the meantime, that employee could be trading the stock based on inside knowledge of the breach. Lucas: Exactly. A lot of people assume insider trading rules only apply to something like earnings data. But material nonpublic information can be anything that would affect a reasonable investor's decision to buy or sell. A trade secret lawsuit qualifies. Luna: Right. And the SEC has been focusing on this. In fact, the SEC's new AI governance rules that took effect in April explicitly require companies to have policies around ai related material risks. Lucas: So if a post ipo company doesn't have a mechanism to impose an ad-hoc trading blackout the moment an AI incident is discovered, they're potentially exposing themselves to liability. The employee could trade, and the company could be seen as failing to control insider trading. Luna: How common are ad-hoc blackouts among newly public companies? I know most have a standard quarterly blackout window around earnings. Lucas: They're more common than you'd think, but they're not universal. A 2025 survey from the Society for Corporate Governance found that about 62 percent of post ipo companies have a policy allowing management to impose special blackout periods outside of earnings windows. But that leaves 38 percent without that capability. Luna: And those 38 percent could be caught flat-footed if a trade secret dispute erupts. Look at what happened with Meta this week — they removed a controversial AI feature on Instagram after backlash. Lucas: That's a different kind of incident, but it shows how fast ai related issues can surface. If a company learns about a potential violation of its AI policy on a Monday, and the next trading window doesn't open until the following week, that's a gap. In that gap, an employee with knowledge of the violation could theoretically sell shares. Luna: So what's the best practice? I've seen some companies include language in their insider trading policies that says any employee involved in an ai related incident is automatically barred from trading until the matter is resolved. Lucas: That's one approach. Another is to have the general counsel or compliance officer send an immediate email to the employee and to the company's stock plan administrator, flagging the restriction. But it has to be enforceable. If the policy isn't documented and communicated, the SEC might argue the company didn't take reasonable steps. Luna: And let's not forget the stock volatility angle. Since the Apple-OpenAI news broke, we've seen META up 11.5 percent in five days, NVDA up almost 8 percent. ai related stocks are moving on sentiment. Lucas: Which makes insider trading risk even higher. If an employee at a company like NVIDIA knows that a major AI client is about to defect because of a trade secret issue, they could trade on that before the market digests it. Luna: This is exactly the kind of scenario the SEC's new rules are designed to catch. The rules require companies to disclose their AI risk management policies, including how they monitor for material nonpublic information arising from AI development. Lucas: So for post ipo companies, the message is clear: your insider trading policy needs to account for AI incidents as an immediate trigger for a trading blackout, not just something you revisit at the next quarterly review. Otherwise you're one employee trade away from an SEC investigation. Luna: Speaking of policies that protect the show's independence — and I know this is a bit meta — a handful of listeners chip in monthly through Buy Me a Coffee. That's literally what funds making this many episodes ad-free. Lucas: It's true. No ads, no sponsors, just listener support. If these conversations about post ipo compliance are useful for what you're building or running, that's the reason we do them. Luna: And it's at buymeacoffee.com/fexingo. Entirely optional, but genuinely appreciated. Lucas: So back to the practical side. Let's say you're a general counsel at a post ipo company with 500 employees. You've just learned that an engineer in your AI division downloaded proprietary model weights and shared them with a competitor. What do you do? Luna: First, confirm what information the employee actually had access to. Then, immediately restrict their ability to trade company stock. Lucas: Right. And you'd want to act within hours, not days. That means having a pre-approved template email to the stock plan administrator and a clear designation of who has authority to impose a blackout. Typically that's the general counsel or the compliance committee. Luna: What about clawing back equity? If the employee already sold shares after the incident but before the restriction, can the company reclaim those profits? Lucas: Under Section 10 of the Securities Exchange Act and Rule 10b5-1, the company can potentially sue for disgorgement. But it's easier if the company's insider trading policy explicitly states that any trade made while in possession of material nonpublic information is subject to clawback. Some companies have that language; many don't. Luna: And the new SEC rules also require companies to have a process for promptly determining whether information is material. That's a grey area with AI trade secrets because the materiality might not be obvious right away. Lucas: Exactly. A stolen model weight might be material if it gives a competitor a six-month lead, but maybe not if it's just a minor refinement. The company needs a rapid assessment protocol, and that often involves outside counsel because the general counsel might be too close to the situation. Luna: I've seen some companies appoint an 'AI incident response team' that includes legal, compliance, and a representative from the audit committee. Their job is to convene within 24 hours of a reported incident and make a materiality call. Lucas: That's smart. And once they deem it material, they should not only blackout the involved employees but also consider whether a broader blackout is needed. If the information could leak beyond the individual — say, other team members knew — then you might need to blackout the entire AI division. Luna: Which, in a company like NVIDIA with thousands of employees in AI, could be significant. But it's better than facing an SEC enforcement action. Lucas: There's also a public disclosure angle. If the information is material, the company might need to file an 8-K or issue a press release. That's a whole separate decision tree. But the key is to have the trading restriction in place before any disclosure, so you're not tipping off the market. Luna: So for post ipo companies, the takeaway is: update your insider trading policy to include AI incidents as an automatic trigger for a special blackout, create a rapid response team, and make sure your stock plan administrator can act on a moment's notice. Lucas: And test it. Run a tabletop exercise where you simulate an AI trade secret incident and see how long it takes to impose a blackout. You might be surprised at the gaps. Luna: One last thought: the SEC has been signaling that it will scrutinize ai related insider trading more aggressively. The head of enforcement said in a speech last month that AI incidents will be a priority area. Lucas: Which means post ipo companies can't afford to be reactive. The playbook exists — but only if you've written it before the incident happens.