For years, the cybersecurity industry has warned that AI would eventually be weaponized against us. In the spring of 2025, it happened — and most small business owners never heard about it.
Researchers at Sysdig documented an attack they named JadePuffer. It wasn’t carried out by a team of hackers in a dark room. There was no human operator making decisions in real time. JadePuffer was an autonomous AI agent — a system that found its way into a target network, figured out what was valuable, encrypted it, destroyed the backups, and left a ransom note. All on its own. From initial access to full encryption, the entire operation ran in minutes.
This isn’t science fiction. It happened. And it changes the conversation we need to have with every small business owner.
What JadePuffer Actually Did
Here’s what happened in plain English.
The AI agent found an unpatched vulnerability in a software tool and walked through the open door. Once inside, it read everything — configuration files, API keys, database credentials, cloud account access. It didn’t just grab what it found; it understood what it found. It prioritized targets, corrected its own mistakes in real time, and moved laterally through the network without a human telling it what to do next.
When a login attempt failed mid-attack, the system diagnosed the problem, adjusted its approach, and executed a corrective sequence — all within 31 seconds. That’s not a hacker at a keyboard. That’s a machine operating faster than any human security team can respond.
It then encrypted 1,342 configuration items using keys that were deliberately made unrecoverable. It dropped the databases. It left a ransom note.
The backups? Gone. The system had been trained on public incident reports and knew exactly where businesses store their recovery options.
Why This Is Different From Every Attack Before It
Traditional ransomware required a human being making decisions at every stage. That created friction — and friction created time. Time for your IT company to notice something unusual. Time for an alert to fire. Time for someone to pull the plug.
JadePuffer eliminated that friction entirely.
The window between a network being breached and being fully encrypted used to be measured in hours or days. Now it can be measured in minutes. Human-speed incident response — calling your IT company, opening a ticket, waiting for a callback — is no longer a viable defense strategy on its own.
There’s a second problem. AI-generated attack code doesn’t look like known malware. Antivirus tools work by recognizing patterns. When the attack code is written fresh by an AI agent for your specific environment, there’s no pattern to recognize. The tools that most small businesses rely on as their primary defense will not catch this.
And here’s the part that should concern every business owner: the AI systems being used to build these attacks were trained on the same public security research, breach reports, and IT documentation that exists on the open internet. They already know that small businesses run outdated software. They already know where the default credentials are stored. They already know that most backup systems aren’t tested.
The Honest Truth About Where Small Businesses Stand
Most of the businesses we work with are in Phase 1 or Phase 2 of the Cerium Lifecycle Blueprint — still building the foundation: MFA, endpoint protection, backup strategy. That work is not wasted. It remains the most important thing you can do.
But JadePuffer makes something clear that we’ve been saying for two years: the foundation is not optional, and the window to build it is closing.
The businesses that get hit hardest by what’s coming are not the ones who got unlucky. They’re the ones who kept waiting.
What You Can Actually Do Right Now
This is not a situation that requires a $150,000 security hire or an enterprise software contract. It requires disciplined execution of the fundamentals — the same fundamentals covered in the $0 Security Stack.
- Patch everything, on a schedule. JadePuffer entered through an unpatched vulnerability. Most small business networks have software running versions that haven’t been updated in months. That’s an open door. Close it.
- Test your backups before you need them. The AI knew where the backups were. The only backups that survive an attack like this are isolated, verified, and tested. If you haven’t run a restore test in the last 90 days, you don’t have a working backup — you have a file.
- Stop using default credentials. Configuration services, network equipment, cloud storage — default usernames and passwords are published on the internet. AI agents know them. Change them.
- Enforce MFA everywhere, not just where it’s convenient. JadePuffer harvested credentials and used them. Multi-factor authentication is the single most effective control against credential-based attacks. Every account, every user, every time.
- Know what’s on your network. The agent found infrastructure by reading configuration files. If you don’t have a current inventory of your systems, software, and third-party integrations, you can’t protect what you don’t know exists.
The Bottom Line
JadePuffer is not the last AI-driven attack. It’s the first documented one. The techniques will get faster, cheaper, and more accessible. What requires significant resources today will be commoditized within 18 months.
The businesses that come through this era of cybersecurity intact are not going to be the ones with the biggest budgets. They’re going to be the ones that built their foundation before they needed it.
That work starts now, and it starts with the basics.
Don’t wait for your own incident report.
The free CyberFortress community is where small business owners build their security foundation — with Phil Turner and a team that’s been watching threats like JadePuffer for years.