Synthetic Minds | Who Holds the Lock When Machines Knock?

Synthetic Minds | Who Holds the Lock When Machines Knock?

A quick note before we start: Synthetic Minds is changing. Instead of daily issues, you will receive one newsletter a week, every Tuesday, with one argument about where technology is taking us and what it means for the people and organizations living through it.

If you are interested in my daily thinking, subscribe to my Substack and LinkedIn, where I share a short take every weekday. Follow me there for the daily signal; your inbox gets the weekly judgment.

And this week I picked up my award in Bangkok for being recognized as the world's #3 futurist, yay! Voting for the 2027 edition is now open, and I would appreciate your vote. Takes ten seconds, and can be done here. On my card, slide How inspirational? to 5, click vote and confirm in your email. Thanks!

All sources are powered by Futurwise.

Your Customers Cannot Hold the Lock Against Machines

Every connected product rests on an assumption nobody wrote down: that the person holding it can keep it safe. I believe it has expired, which turns security from a line in the IT budget into a question for the CEO and the board.

A German security firm logged in to the cloud Sungrow uses to manage its solar systems, and the login ignored the password. The researchers put its reach at gigawatts; the homeowners behind it believed the lock was theirs.

Most of our systems were built by humans, for humans, and we keep bolting new technology onto them without rebuilding the foundations. Yet the ones probing them are increasingly machines. An AI agent found a flaw in the XRP Ledger, a payment network that had survived 11 years of human review, and built a working attack to prove it.

The test of time is the test of the machines

No team of people can prevent every vulnerability, so building something safe means putting the machines that could breach it on your side, testing your code before someone else's machines do. The XRP case shows it working: a defender found the flaw, and the fix shipped three days later.

Security can no longer be an afterthought, because a business whose security is not in order will eventually lose to the machines that find its flaws, and passing their test takes far more time, energy and money than it did before AI took over hacking. Which of your products was designed for an attacker who gets tired?

Leaving the lock with the customer is already failing on the grid. Millions of homes with solar panels and batteries, pooled by software into virtual power plants, make it more flexible, which I welcome. But they push security into homes where few people know how to secure an appliance, so the responsibility shifts to installers and the companies running those plants. When one control dashboard can be shown steering 10 GW across 1 million customers, a grid this flexible is also open to agent swarms: AI agents probing every door at once.

Nobody signed up to carry this

Smart glasses raise the same question. Meta has opened the camera in its glasses to outside developers, so apps the wearer never wrote will increasingly decide what it records. Europe's privacy regulators say makers and app providers cannot pass their duties to the wearer. It should be law rather than guidance, because big tech will not govern itself.

I felt this at an event, when a man kept taking his glasses off and on, and I ended the conversation because I could not tell whether he was recording. Once that moment is everywhere, people stop trusting whoever wears one, friends and family included. Those who never bought a pair have one control left: ending the conversation, or the friendship. Let's not give The Zuck the pleasure of affecting your friendships.

A company that quietly hands its security, or other people's privacy, to someone who never agreed to carry it does not deserve to exist, and customers should walk away, however convenient the service.

One more thing: the quantum date is not yours to borrow

Quantum computers will break the encryption that protects us, though nobody knows when. AI is moving faster: within 16 days, AI agents helped set two public records for cracking encryption challenges, one with Devin and one with Claude. Standard keys remain safe; the shorter keys in older systems cost roughly $30 million each to crack. Big tech says what benefits big tech, so do your own research.

The question for your next leadership meeting

What is a product worth when its customers have to secure it themselves?

Security and privacy are not a nice to have. If we treat them as rights, that comes with consequences.

If your executives sit down together before the next budget round, the Intelligence Age Leadership Briefing gives them the room to decide which risks they have left with the people they serve.

If this newsletter was forwarded to you, you can sign up here.

Thank you.
Mark