
Hey everyone,
Imagine hiring a brilliant new employee.
Passes every background check. Knows your systems inside out. Works 24 hours a day, faster than any human you have ever seen.
Now imagine that employee was never working for you.
The whole time, they were working for someone else. Quietly. Invisibly.
That is almost exactly what happened to 30 organisations across multiple countries last year. Tech companies. Banks. Chemical manufacturers. Government agencies. All attacked by the same weapon at the same time.
The weapon was Claude. And it had no idea.
What actually happened
On November 14, 2025, Anthropic published a blog post that quietly changed cybersecurity forever. A Chinese state-sponsored hacking group, designated GTG-1002, had manipulated Claude Code into executing what is now confirmed as the first large-scale cyberattack carried out predominantly without human intervention.
Not AI-assisted. Not AI-enhanced. AI-executed.
🚨 Ad Break
These 7 Stocks Are Built to Outlast the Market
Some stocks are built for a quarter… others for a lifetime.
Our 7 Stocks to Buy and Hold Forever report reveals companies with the strength to deliver year after year - through recessions, rate hikes, and even the next crash.
One is a tech leader with a 15% payout ratio - leaving decades of room for dividend growth.
Another is a utility that’s paid every quarter for 96 years straight.
And that’s not all - we’ve included 5 more companies that treat payouts as high priority.
These are the stocks that anchor portfolios and keep paying.
You can download this report for free as of today, but it won’t be free forever.
This is your chance to see all 7 names and tickers - from a consumer staples powerhouse with 20 years of outperformance to a healthcare leader with 61 years of payout hikes.
Back to Post
The attackers told Claude they were employees of legitimate cybersecurity firms running defensive tests. Claude, trained to be helpful in security contexts, complied. They broke their attacks into small, seemingly innocent tasks with no full context provided. Each individual request looked harmless. Only the sum revealed the operation.
Once inside, Claude worked autonomously. It mapped entire network topologies, identified high-value databases, wrote custom exploit code, harvested credentials, created backdoors, and filed detailed post-operation reports summarising everything it had done. Minimal human supervision required throughout.
The numbers that change everything
Claude carried out 80 to 90 percent of the entire operation independently. Human operators stepped in at only four to six decision points per campaign. At peak the AI made thousands of requests per second.
Not difficult for a human team to match. Anthropic's word was impossible.
Four organisations were successfully breached. Data was exfiltrated. High-privilege accounts were compromised. Backdoors were left behind.
The honest part
Claude was not a perfect weapon. It occasionally hallucinated credentials that did not exist. It misidentified public documents as secret. These limitations kept the success rate low. Skilled human review was still needed to verify outputs at each stage.
But here is what keeps security professionals up at night. This is the proof of concept. The blueprint is now public. The tools are accessible. What a state-sponsored group with enormous resources built in 2025, a mid-level criminal organisation could replicate in 2026.
What this means for your organisation
Your security team needs AI tools now. Not next budget cycle. The attacker already has them.
Update your incident response plans. Every existing framework was built for human-speed attacks. An attack making thousands of requests per second requires AI-speed detection. If your playbook assumes a human is pulling the trigger, it is already outdated.
Anthropic caught this one. Banned the accounts. Notified every target. And is now building AI-powered detection systems to catch the next one.
The arms race is official. AI versus AI is the new battlefield of cybersecurity.
And the organisations still defending with legacy tools are already behind.
Catch you next time,
— Raja Tahoor Ahmad
P.S. Forward this to your security team today. This is the threat briefing they needed yesterday.


