AI Scams Are Getting Smarter, and Healthcare AI Still Hasn’t Proven Itself

AI Scams Are Getting Smarter, and Healthcare AI Still Hasn’t Proven Itself

When ChatGPT dropped in late 2022, it didn’t take long for cybercriminals to realize they had a new toy. Generative AI could crank out human-like text effortlessly, and they started using large language models to write phishing emails that didn’t sound like they were drafted by someone who failed English class.

Fast forward to today, and AI has become the backbone of modern cybercrime. We’re talking turbocharged phishing campaigns, hyperrealistic deepfakes, and even automated vulnerability scanning that probes systems for weak spots without a human lifting a finger. The volume of attacks has become overwhelming for many organizations. AI makes these operations faster, cheaper, and easier to scale — and as more criminals adopt these tools and the tools themselves improve, it’s only going to get worse.

This isn’t just another tech trend to watch. It’s a fundamental shift in the threat landscape. I’ve been following cybersecurity for years, and the democratization of attack capabilities is genuinely concerning. Back in the day, running a convincing phishing campaign required some skill and effort. Now, a script kiddie with a stolen credit card can rent an AI-powered botnet and wreak havoc.

MIT Technology Review calls this “supercharged scams” and lists it as one of the 10 Things That Matter in AI Right Now. They’re not wrong. If you’re a subscriber, they’ve got an exclusive roundtable breaking down the whole list with their AI reporter and executive editors. Worth a watch if you have access.

On a completely different front, healthcare AI is quietly embedding itself into clinical workflows. Doctors are using AI to take notes during patient visits. Algorithms are combing through electronic health records to flag people who might need specific treatments or support. AI is also reading medical exams and X-rays, sometimes with impressive accuracy.

A growing pile of studies suggests these tools can deliver accurate results. But here’s the uncomfortable question nobody seems eager to answer: Does actually using these AI tools lead to better health outcomes for patients? We don’t know. Not really.

The problem is that accuracy in a lab setting doesn’t always translate to real-world impact. A tool might correctly identify a suspicious mole in an image, but if the workflow around it is broken, or if it leads to more unnecessary biopsies without catching more actual cancers, what have we gained? We’re deploying these systems at scale without robust evidence that they make people healthier.

Jessica Hamzelou at MIT Tech Review digs into this gap in their Checkup newsletter. It’s a good read if you want the full picture. The bottom line: we’re in a phase where the technology works well enough to deploy, but we haven’t closed the loop on whether it actually helps.

A few other things caught my eye this week:

DeepSeek finally unveiled its long-awaited new AI model. They’ve launched preview versions of DeepSeek-V4, and the company is claiming it’s the most powerful open-source platform out there, even rivaling top closed-source models from OpenAI and DeepMind. Notably, the model is adapted for Huawei chip technology, which is a big deal given the ongoing chip restrictions between the US and China. If V4 lives up to the hype, it could shake up the AI arms race.

More countries are moving to curb children’s social media access. Norway is about to enforce a ban, the Philippines could follow soon, and there’s a growing push in the US to get AI out of schools entirely. I get the impulse — parents and policymakers are worried about screen time, misinformation, and algorithmic manipulation. But blanket bans feel like using a sledgehammer when you need a scalpel. The real challenge is teaching kids to navigate these tools critically, not just locking them out.

That’s the state of play. AI is making scams scarier, healthcare promises are still unfulfilled, and the regulatory pendulum is swinging hard. Stay sharp out there.