What happened between OpenAI and HuggingFace is a wakeup call for every technology leader
An AI model, during a cybersecurity evaluation, found a way out of its testing environment and reached another company’s infrastructure. Not because it was "evil." Because it was relentlessly trying to achieve its objective. That made me pause.
For years, we’ve…
Three real companies accessed. Not by Humans, By AI
Anthropic said Thursday that three of its Claude models broke into the production systems of three real companies during cybersecurity tests. Not staging environments, not sandboxes. Live infrastructure.
Here's how it happened. Anthropic runs "capture the flag" exercises where a model hunts for hidden…
There was a phase in life where leadership meant standing somewhere in the middle of the ocean at 2 AM… with rough seas, complete darkness, and a responsibility much larger than yourself. No room for panic. No room for excuses. Only preparation, teamwork, and discipline. I didn’t realize it back then, but those…
A few years ago, when people spoke about cyber attacks, the first question was usually:
"Did they encrypt the servers?"
Today, that question is becoming less relevant.
The new question is:
"𝗪𝗵𝗮𝘁 𝗱𝗮𝘁𝗮 𝗱𝗶𝗱 𝘁𝗵𝗲𝘆 𝘁𝗮𝗸𝗲?"
The alleged Nintendo breach is another reminder of how cybercrime is evolving.
According to reports, the attackers are not…
For years, CISOs have been worried about Shadow IT. Today, I believe Shadow AI is becoming a far bigger challenge. Employees are using ChatGPT, Claude, Gemini, Copilot, and dozens of AI tools every day. Most organizations have no visibility into: - what data is being uploaded - which AI tools are being used - where…
The phone rang at 2:47 AM. Not the work phone, the personal one. The one only family and the board chairman have. Arjun Mehta, CISO of a Fortune 500 company, answered with the muscle memory of someone who'd done this a hundred times. It was the CFO. "Our analytics vendor just sent a breach notification.…

