A customer service chatbot confidently describes a product that doesn’t exist. A financial AI invents market data. A healthcare bot provides dangerous medical advice. These AI hallucinations, once dismissed as amusing quirks, have become million-dollar problems for companies rushing to deploy artificial intelligence.

Today, Patronus AI, a San Francisco startup that recently secured $17 million in Series A funding, launched what it calls the first self-serve platform to detect and prevent AI failures in real-time. Think of it as a sophisticated spell-checker for AI systems, catching errors before they reach users.

Inside the AI safety net: How it works

“Many companies are grappling with AI failures in production, facing issues like hallucinations, security vulnerabilities, and unpredictable behavior,” said Anand Kannappan, Patronus AI’s CEO, in an interview with VentureBeat. The stakes are high: Recent research by the company found that leading AI models like GPT-4 reproduce copyrighted content 44% of the time when prompted, while even advanced models generate unsafe responses in over 20% of basic safety tests.

The timing couldn’t be more critical. As companies rush to implement generative AI capabilities — from customer service chatbots to content generation systems — they’re discovering that existing safety measures fall short. Current evaluation tools like Meta’s LlamaGuard perform below 50% accuracy, making them little better than a coin flip.

Patronus AI’s solution introduces several innovations that could reshape how businesses deploy AI. Perhaps most significant is its “judge evaluators” feature, which allows companies to create custom rules in plain English.

“You can customize evaluation to exactly [meet] your product needs,” Varun Joshi, Patronus AI’s product lead, told VentureBeat. “We let customers write out in English what they want to evaluate and check for.” A financial services company might specify rules about regulatory compliance, while a healthcare provider could focus on patient privacy and medical accuracy.

From detection to prevention: The technical breakthrough

The system’s cornerstone is Lynx, a breakthrough hallucination detection model that outperforms GPT-4 by 8.3% in detecting medical inaccuracies. The platform operates at two speeds: a quick-response version for real-time monitoring and a more thorough version for deeper analysis. “The small versions can be used for real-time guardrails, and the large ones might be more appropriate for offline analysis,” Joshi told VentureBeat.

Beyond traditional error checking, the company has developed specialized tools like CopyrightCatcher, which detects when AI systems reproduce protected content, and FinanceBench, the industry’s first benchmark for evaluating AI performance on financial questions. These tools work in concert with Lynx to provide comprehensive coverage against AI failures.

Beyond simple guard rails: Reshaping AI safety

The company has adopted a pay-as-you-go pricing model, starting at $10 per 1000 API calls for smaller evaluators and $20 per 1000 API calls for larger ones. This pricing structure could dramatically increase access to AI safety tools, making them available to startups and smaller businesses that previously couldn’t afford sophisticated AI monitoring.

Early adoption suggests major enterprises see AI safety as a critical investment, not just a nice-to-have feature. The company has already attracted clients including HP, AngelList, and Pearson, along with partnerships with tech giants like Nvidia, MongoDB, and IBM.

What sets Patronus AI apart is its focus on improvement rather than just detection. “We can actually highlight the span of the specific piece of text where the hallucination is,” Kannappan explained. This precision allows engineers to quickly identify and fix problems, rather than just knowing something went wrong.

The race against AI hallucinations

The launch comes at a pivotal moment in AI development. As large language models like GPT-4 and Claude become more powerful and widely used, the risks of AI failures grow correspondingly larger. A hallucinating AI system could expose companies to legal liability, damage customer trust, or worse.

Recent regulatory moves, including President Biden’s AI executive order and the EU’s AI Act, suggest that companies will soon face legal requirements to ensure their AI systems are safe and reliable. Tools like Patronus AI’s platform could become essential for compliance.

“Good evaluation is not just protecting against a bad outcome — it’s deeply about improving your models and improving your products,” Joshi emphasizes. This philosophy reflects a maturing approach to AI safety, moving from simple guard rails to continuous improvement.

The real test for Patronus AI isn’t just catching mistakes — it will be keeping pace with AI’s breakneck evolution. As language models grow more sophisticated, their hallucinations may become harder to spot, like finding increasingly convincing forgeries.

The stakes couldn’t be higher. Every time an AI system invents facts, recommends dangerous treatments, or generates copyrighted content, it erodes the trust these tools need to transform business. Without reliable guardrails, the AI revolution risks stumbling before it truly begins.

In the end, it’s a simple truth: If artificial intelligence can’t stop making things up, it may be humans who end up paying the price.

Getty Images

Getty Images

Getty Images

What was agreed at the summit?

Bonobos founder Andy Dunn is back in the builder’s seat, working on an in-person social media platform called Pie. But the biggest lessons he learned from his $310 million Bonobos exit don’t have as much to do with entrepreneurship as they do with staying sane.

When Dunn was in college, he was diagnosed with bipolar disorder, but he didn’t get adequate treatment until 2016, when he was hospitalized during a manic episode for the second time.

“The manic state is just a disaster — that’s like being in psychosis, you know, messianic delusions. … You can’t accomplish anything in that state,” Dunn said onstage at TechCrunch Disrupt 2024. The incident was enough of a wakeup call that 16 years after his initial diagnosis, he finally took his condition seriously and started going to therapy, taking medication, and monitoring his sleep.

Dunn wrote a book called “Burn Rate: Launching a Startup and Losing My Mind,” documenting the parallel processes of building Bonobos and figuring out how to accept and then manage his bipolar disorder. But the lessons from the book are applicable for entrepreneurs beyond those with Dunn’s diagnosis.

“We all have mental health, right? It doesn’t take a diagnosis to suffer or struggle,” he said.

Still, entrepreneurs tend to report a higher incidence of mental health issues throughout their lives than the average person.

“There’s definitely a correlation between neurodivergence and creativity,” he said. “I don’t know if entrepreneurship attracts people who are neurodivergent, or it makes them more neurodivergent, but there’s certainly some kind of a virtuous and sometimes unvirtuous cycle there.”

That interplay between mental illness and entrepreneurship is even more palpable for Dunn, who says that the state of hypomania — the high of bipolar disorder, as opposed to the crushing depressive periods — could be conducive to running a startup.

“Here are the DSM criteria for [hypomania]: rapid speech, increased ideation, grandiosity, decreased need for sleep, ability to be more creative … more or less the central casting traits of an entrepreneur having a good day,” he said. “I was able to benefit from that, but the price that I paid was ultimately too high. I was depressed with suicidal ideation for between two to three months a year, and then ultimately, the full mania and psychosis came raging back, which was catastrophic.”

But even in an astonishingly productive hypomanic state, Dunn doesn’t think he was the greatest boss or colleague. He said that one of the side effects of hypomania is becoming irritable when people disagree with you, which is essential to running a collaborative company. Now, running Pie, Dunn welcomes this debate.

“When we disagree, let’s go, let’s disagree even more, because we’re going to be able to make a better decision coming out of it,” he said.

While discussions about mental health have become more mainstream, founders still worry about the stigma of revealing a diagnosis to colleagues and investors. Dunn is an adviser to the Founder Mental Health Pledge, which asks investors to advocate for the mental health of the founders they invest in. But he’s not naive that the stigma is still present — when founders ask for his advice about when to disclose a mental health concern to investors, he says to wait six weeks until after the deal closes.

“We raised $125 million at Bonobos — would you give $125 million to someone who can either be psychotic or catatonically depressed?” Dunn said. “But also, you shouldn’t do what I did and hide it, because then, you know, when there is a crisis, it’s a surprise.”

Dunn’s discussion of his experience with bipolar disorder doesn’t seem to have hurt his ability to fundraise, though — Pie just raised a $11.5 million Series A. As public as he is about the severity of bipolar disorder, he’s also open about how his regimen of therapy and medication have helped him live a stable life.

“I treat bipolar as my Olympic regimen. For Simone Biles, it’s how to navigate and win the gold,” he said. “For me, the gold medal is to die of something else, right? Because the horrible thing about bipolar is the suicide rate.”

Now, the next test for Dunn is to do the work it takes to make Pie a success without sacrificing his stability.

“Here’s the challenge,” Dunn said. “We want to have good mental health, and we want our teams to have balance in mental health, and yet a 40-hour workweek doesn’t cut it. You can’t change the world with a bunch of people working 40 hours a week.”

One way Dunn has navigated this fine line is to be open with job candidates about what the work will entail, as well as how he will support them with company benefits.

“I have a new spiel I give when recruiting, which is, this is a 50- to 60-hour-per-week job, and in return, you’re going to get two awesome things. One, you’re going to learn more and grow more and develop more. Two, you’ve got equity,” he said.

Like any startup leader, Dunn wants his team to work hard, but he believes there’s a way to do that without it backfiring. In describing his time at Bonobos in “Burn Rate,” Dunn writes, “I came to a classic mistaken conclusion of an immature startup founder: if the business isn’t working, then we must not be working hard enough.”

There’s no denying that founders need to work hard — but taking care of oneself is part of that hard work.

A new report from Heimdal has revealed jobseekers across the world are being targeted by scams exploiting individuals looking for work in sectors such as finance, IT, and healthcare.

Based on an analysis of over 2,670 social media posts and comments from victims in 2023 and 2024, the report highlights the common tactics used by scammers, the industries most affected, and the emotional toll these scams take on their victims.

This week, Apple announced a new M4 iMac. It got some upgrades that help make it more appealing to creatives and pros, such as the more powerful M4 chip, Thunderbolt 4, upgraded camera, and nano-texture display.

But an iMac Pro, this is not.

A more larger and more powerful iMac has been missing in the lineup since before the transition to Apple Silicon. Despite Apple’s insistence that it completed the transition to Apple Silicon in 2022, a 27-inch iMac is still missing. Is there any hope?

Has it been replaced?

Yet conspicuous by its absence was any kind of larger iMac equipped with an M4 chip. Close to four years after the iMac Pro was discontinued, we’re still left wondering what — if anything — Apple plans for its plus-sized all-in-one computer.

And because of that, I can’t help feeling that there’s a noticeable gap in Apple’s desktop Mac lineup, whether we’re talking about a beefy iMac Pro or simply a larger iMac with the same colorful design as Apple’s base model. But will Apple actually do anything about this situation?

For its part, Apple has said that it thinks the combination of the Mac Studio and the Studio Display is a great combination for professional users. And on that point, it might be right: together, the Mac Studio and Studio Display start at $3,598, which is much cheaper than the iMac Pro’s old starting price of $4,999. Sure, the outcome is not quite as sleek, but it does a great job for almost $1,500 less. Who can argue with that?

In that regard, I’m inclined to agree that a true iMac Pro probably doesn’t need to be brought back, and that Apple was right to keep it out of the recent Mac announcements. It feels like more of a “nice to have” than an absolute essential, and I doubt that’s enough justification for Apple.

But what about the larger desktop iMac with an M4 Pro? Apple used to sell the iMac in both 21.5-inch and 27-inch sizes, yet now your only choice is the 24-inch iMac. For people who want a larger all-in-one computer without needing the beefier components (or hefty price tag) of the Mac Studio and Studio Display, there are few options from Apple.

Missing in action

Over the past few months, the idea that Apple was planning to release a bigger iMac in the fall simply hasn’t been on the agenda. In fact, there have been precious few rumors surrounding this mooted device at all in the last year or so. To me, that implies either that a larger iMac is a long way off, or that it’s not happening at all.

Even the upgrades seen by the actually existing iMac are minor at the moment. Sure, this year we’ve had a few new colors, a Thunderbolt 4 upgrade and, yes, the M4 chip, but it’s not exactly a radical overhaul of the iMac. Does this computer simply not sell well enough for Apple to devote significant resources to it? If so, that could explain why we’re not seeing a larger model — perhaps Apple just doesn’t think an upgrade like that is worth anyone’s time.

Advertisement

If so, that’s a sad state of affairs for such an iconic computer, but such is the world of technology. Things move on at such a rapid pace that it’s inevitable that some once-popular devices will get left behind.

The other possibility is that Apple is merely biding its time until it feels the larger iMac is ready before releasing it. The iMac doesn’t sell in huge numbers like iPhones and MacBooks, so it’s less important for Apple to get new updates out of the door every single year. We’ve already heard that Apple is still “exploring” the idea of a larger iMac, with Bloomberg journalist Mark Gurman previously stating that Apple is still working on the product. When the timeframes are longer, Apple can afford to move at a slower pace.

I can’t tell you for sure whether Apple is ever going to bring back an iMac with a bigger screen, or what form that will take if it does happen. But what is certain is that we’ve got at least another year ahead of us before we see the next iMac upgrade. Let’s hope that one comes with some better news for fans of larger displays.