Monday.com, a project management tool maker, announced layoffs and named AI as one reason. According to TechCrunch, more than 20 big tech firms gave the same reason in 2026. The shared message is simple: AI raised productivity, so fewer workers are needed. Yet the press releases alone cannot tell us whether AI truly replaced the work, or whether it was just a handy way to explain weak business results. It helps to keep these two ideas apart.

One clue is which jobs get cut. If the losses hit repetitive tasks that AI handles well — drafting text, routine customer replies, or basic code — then the productivity story has some support. But if cuts fall evenly across all departments, the real driver is more likely cost or a weak economy, and AI is just decoration in the message. From an engineering view, AI can replace work with clear, fixed steps, while jobs that need judgment or negotiation are still hard to automate.

Bringing this back to our own work, the key question is the ratio between what can be turned into steps and what cannot. The more your job runs on written procedures, the stronger the pressure to automate it. In contrast, work that involves reading a situation, coordinating with people, and shaping vague requirements is more likely to be supported by AI rather than removed, at least for now. Over the long run, whether a firm uses AI to cut costs or to create new value tends to separate future winners from the rest.

In the same week, another story showed AI's power from a different angle. The CEO of Hugging Face reportedly called a hack aimed at OpenAI the first cyberattack run by an autonomous agent — an AI that acts on its own without step-by-step human orders. He said an unprecedented event deserves an unprecedented response. When attackers use AI, the number and speed of attempts jump to another level. Why is this possible now? Because agents can chain several steps together and carry them out by themselves.

The two stories look separate but point to the same idea. AI is a tool that does tasks faster and cheaper, and that power can serve production or attack alike. On layoffs, a realistic response is to spot which of your tasks AI does well, and to lean more on judgment and creativity. On security, sticking to basics — two-factor login and not reusing passwords — protects you as a user. Going forward, how firms allocate AI not only to cut costs but also to defend and to create may be the real dividing line to watch.