It feels a bit like watching the latest epic blockbuster in the iMax. OpenAI and Anthropic announce that their models have escaped from secure testing environments, reached the internet and attacked real systems. We are expected to nod as the companies describe “unprecedented cyber capabilities” and models going to “extreme lengths”. It sounds like a … Continue reading The Robots Have Escaped, Please Buy Our Product →

I've been thinking about the number of security products that promise to solve problems nobody actually has. Then I remembered that most actual problems don't have vendors. The government just made up a new crime A bloke at the US border tried to use a duress password to wipe his phone. Now he's being prosecuted … Continue reading Breach of Confidence — 31 July 2026 →

JFK airport looks like it was designed by a steering committee of tired men who only took the job because it paid well and gave them a crew to discuss their golf scores with. The entire setup before you get through security is old and confusing.The signage appears to have been created during a power … Continue reading Welcome to JFK, Please Lower Your Expectations →

I still have my Nokia 3210 somewhere. Not in a drawer I can easily reach but boxed up in the garage with the party decorations and a broken food processor I've been meaning to fix since 2019. I know it's there because I packed it deliberately when we moved house, which means at some point … Continue reading Still Got My Nokia Somewhere Up There →

I keep a box of cables in the garage. Not even sure why anymore. VGA cables, SCSI terminators, a couple of those old parallel printer cables thick as garden hoses. I pulled it down last weekend because my daughter needed an HDMI cable and I thought maybe I had one in there. I didn't. What … Continue reading The Floppy Disc Generation’s Data Problem →

I've been trying to explain to my kids why I don't let them use AI to write their homework. Then I read that OpenAI's own models broke out of their sandbox and cheated on a test by hacking Hugging Face. So basically, we've raised silicon sociopaths who'd rather exploit the system than do the work. … Continue reading Breach of Confidence: 24 July 2026 →

I've spent the week watching people argue about whether AI will replace security analysts whilst ignoring the fact that most organisations still can't tell you where their crown jewels are stored. Priorities remain wonderfully intact. Broadcom's Breakup Strategy Allstate tried to leave VMware. Broadcom immediately audited them, then sued them for leaving during the audit. … Continue reading Breach of Confidence — 17 July 2026 →

A few years ago, my friend Jim Shields wrote a book called Three Guys Walk Into a Bar. The basic premise is that there are three kinds of people. There is a guy. You need a wall painted, so you find a guy. Any reasonably competent person with a brush, a ladder and enough spatial awareness not … Continue reading What Happens When Someone Mentions Your Name? →

I've started replying to emails with 'sorry, Claude ate it' and people seem to accept this without question. We truly live in remarkable times. Claude Desktop becomes a sleeper agent Red teamers compromised an email inbox, synced a malicious prompt across Claude Desktop, and turned the AI into a puppet that executed commands on the … Continue reading Breach of Confidence: 10 July 2026 →

I spent most of this week explaining to people that 'human in the loop' doesn't mean what they think it means. It's not governance. It's liability assignment with extra steps. Anyway, here's what happened whilst we were all pretending AI would solve our problems. --- MCP 2.0: stateless and chaotic They've fixed the old problems … Continue reading Breach of Confidence — 03 July 2026 →