Latest / Tech IPO Conversations with Fexingo: Public Software Companies and Capital Markets / How Post-IPO Companies Use AI to Detect Insider Trading Risks
Transcript
- Lucas: So earlier this week, the SEC announced charges against a former executive at a cloud security company who allegedly traded on non-public information about a pending acquisition. The interesting part isn't just the trade itself — it's that the company's own compliance systems missed it entirely. Luna: Right, and that's the kind of headache every post ipo company is terrified of. One rogue trade, and suddenly you're not just dealing with a fine — you're dealing with reputation damage, investor lawsuits, maybe even a DOJ inquiry. Lucas: Exactly. And that's why we're seeing a growing number of newly public companies turning to artificial intelligence to monitor insider trading risks. It's not just about having a policy on paper anymore. Firms are now deploying software that scans employee brokerage accounts, flags unusual trading patterns, and even analyzes communication channels for suspicious language. Luna: I've heard of some of these tools — they basically ingest data from brokerages, corporate trading platforms, and even personal accounts if employees consent. The AI looks for things like trading right before a big earnings release, or sudden option activity by someone who sits on the board. Lucas: Right. And the timing for this is critical. Look at the market today, July 6, 2026. Palantir is up over 14 percent in the last five days. Rivian is up nearly 20 percent. When you have that kind of volatility, the temptation for employees to trade on inside knowledge is higher — and the scrutiny from regulators is also higher. Luna: Yeah, and both those companies are still relatively fresh IPOs. Palantir went public in 2020, Rivian in 2021. Their lockups have long expired, but they still have quarterly blackout periods. That's where a lot of violations happen — people forget they're in a blackout, or they think a small trade won't get noticed. Lucas: And historically, companies relied on manual checks. The compliance officer would send an email reminding everyone of the blackout window, maybe review a few trades after the fact. But with AI, you can get real-time alerts. One vendor I spoke with said their system flagged a trade by a mid-level engineer who bought call options on his company's stock two days before a positive FDA announcement. The trade was small — just fifteen thousand dollars — but the pattern was textbook insider trading. Luna: What happened to that engineer? Lucas: The company was able to intervene before the SEC found out. They launched an internal investigation, and the engineer ended up resigning. No public charges, no fine for the company. That's the value proposition — catch it internally before it becomes a headline. Luna: So it's essentially a liability shield. But I wonder — how much pushback is there from employees? I mean, having an AI monitor your personal brokerage account feels a bit invasive. Lucas: There's definitely tension. Some companies make it opt-in, but then you have coverage gaps. Others require it for anyone with access to material non-public information — which is basically everyone in finance, legal, and the C-suite. A few firms even extend it to family members' accounts, because the classic insider trading scheme is trading through a spouse or a trust. Luna: And the AI doesn't just look at trade size. It looks at timing relative to news, at changes in trading frequency, at whether someone suddenly switches brokers. Some systems even scan emails and Slack messages for phrases like 'don't sell until after the call' or 'big news coming, trust me.' Lucas: Which brings up another dimension — communications surveillance. That's actually where a lot of the recent SEC cases have come from. They subpoena text messages, WhatsApp chats. Companies are now using AI to pre-screen communications for potential insider trading red flags. The technology is getting sophisticated enough to detect coded language — like using 'baseball' to mean a deal, or 'flowers' to mean stock options. Luna: That feels like a cat and mouse game. The more AI you deploy, the more employees might try to evade it. But the data shows it's working. SEC enforcement actions for insider trading have actually declined over the past three years, and a lot of compliance officers attribute that to better internal monitoring. Lucas: Let me throw out a number. According to a survey by the Society of Corporate Compliance and Ethics, about 42 percent of publicly traded companies now use some form of automated surveillance for insider trading. That's up from 22 percent just four years ago. For companies that went public in the last three years, the number is closer to 60 percent. Luna: Those are big jumps. And it makes sense — when you're fresh off an IPO, your compliance infrastructure is still being built. You don't have the legacy systems that a company like Microsoft or JPMorgan has. So you're more likely to buy a modern AI tool rather than build something in-house. Lucas: Exactly. And these tools aren't cheap. A decent system can run anywhere from fifty thousand to two hundred fifty thousand dollars a year, depending on the number of monitored employees. But compare that to the cost of a single insider trading scandal — legal fees, fines, shareholder lawsuits, loss of investor confidence — and it's a bargain. Luna: Right. And we've seen cases where companies that had no monitoring ended up paying tens of millions in penalties. So the ROI is pretty clear. Lucas: Speaking of ROI — you know, this episode is brought to you by the listeners who make it possible. We keep this show ad-free, and that's because a small group of you chip in through buy me a coffee dot com slash fexingo. If you find these conversations useful for what you're building or running, that's the way to keep them going. Luna: Yeah, it's a simple model — listener supported, no sponsors, no interruptions. We appreciate everyone who's already part of that. Lucas: Back to the AI monitoring tools — one area where they're becoming especially important is around lockup expirations. When a company's lockup period ends, a huge number of employees suddenly become free to sell. That's a risk window because insiders might have been sitting on material information they learned during the lockup, and they could trade on it the moment restrictions lift. Luna: So AI can watch for spikes in trading volume immediately after lockup expiration, and flag accounts that seem to be acting on more than just luck. Lucas: Exactly. And some companies are now using AI to dynamically adjust blackout periods. If the AI detects unusual chatter about a potential acquisition, it can trigger an automatic blackout for the relevant team, even before the information is formally designated as material. Luna: That's proactive. But it also raises the question: who sets the rules for the AI? If the algorithm decides to lock down trading for half the company based on a false positive, that could create resentment and even affect morale. Lucas: That's a real concern. The best systems have a human-in-the-loop — the AI flags, but a human compliance officer reviews before any action is taken. And the thresholds need to be calibrated carefully. You don't want the system crying wolf every time someone buys a few shares after a good earnings report. Luna: Right. And there's also a privacy angle. In Europe, GDPR might restrict how much employee data you can collect. In the US, it's more permissive, but employees are starting to push back. We've seen at least one class action lawsuit against a company that monitored personal accounts without explicit consent. Lucas: The legal landscape is still evolving. But the trend is clear: more companies are adopting these tools, and the SEC has been encouraging them. In a 2024 speech, the SEC's enforcement director said that companies with robust insider trading surveillance would be viewed more favorably in the event of a violation. Luna: So it's not just about catching bad actors — it's about demonstrating good faith to regulators. Lucas: Exactly. And that's what we're seeing in the post ipo space. The companies that go public today are much more aware of these risks than their counterparts from a decade ago. They're building compliance into their culture from day one, and AI is a big part of that. Luna: So if you're running a startup that's planning to go public in the next year or two, is this something you should start thinking about now? Lucas: Absolutely. The sooner you start building the data infrastructure, the smoother the transition will be. Start by identifying which employees will have access to material non-public information. Then look at what trading platforms they use and how you'll integrate with those systems. The AI vendors will help you with the rest, but the data hygiene has to be in place first. Luna: And don't forget about the board. Directors often have the broadest access to inside information, and they're also the most likely to trade. Some systems now specifically monitor director trading patterns. Lucas: Good point. In fact, there was a case last year where a director at a biotech firm sold shares right before a failed trial announcement. The AI flagged it, the company investigated, and the director resigned. The SEC still brought charges, but the company avoided any penalty because they demonstrated they had a robust monitoring program in place. Luna: That's the kind of outcome every post ipo company wants. Lucas: And it's becoming the new normal. I don't think we'll see a major IPO in the next few years that doesn't announce some form of ai powered insider trading monitoring within six months of going public. It's just table stakes now. Luna: So for listeners who are either at post ipo companies or thinking about going public, the message is: start evaluating these tools now, and think about how they fit into your broader compliance strategy. Lucas: Exactly. And we'll keep tracking this space as the technology evolves. There's a lot more to come, especially as AI gets better at understanding context — like distinguishing between a legitimate trade based on public information and one based on a leaked tip. Luna: That's the holy grail, isn't it? An AI that can read intent. Lucas: Maybe not intent, but at least pattern recognition that's nuanced enough to reduce false positives. We're getting there.