Grand Theft Auto made crime a game. AI is doing the same for hacking
· Fortune

Fans have waited 13 years for Grand Theft Auto VI. The wait for GTA V’s sequel dragged on through so many delays that it became a meme. When a game takes that long, you’d expect it to feel dated on arrival. But when GTA VI finally arrived this week, the opposite thing happened. The real world has started to look a lot like the game.
Six days before the launch, the FBI issued a FLASH warning about ARTEX, an open-source tool that lets AI agents carry out an intrusion “at machine speed, from reconnaissance through data exfiltration.” Some attacks reached privileged access in under three hours. The GitHub repository has since been taken down, but copies are still out there, and it has already been used against banks, government agencies, and tech companies. Skill, time, and money used to separate a bored teenager from a professional crew. Now a bored teenager with AI can operate like one.
Visit betsport24.es for more information.
Rockstar knows what a bored teenager can do. In 2022, a 17-year-old broke into the company and leaked early GTA VI footage from a hotel room, using an Amazon Fire Stick and a phone. That took some serious skills and marked the last time anyone would need them. Rockstar spent 13 years building a game that feels like real life. In the same stretch, real life took on the rules of the game. AI turned the hacking craft into a criminal loop. You pick a target, try, fail, try again, at almost no cost, with the push of a button, from your couch. While developing the latest GTA took more than a decade, reality caught up in a fraction of that.
That shift left defenses behind, because they were built for a slower pace. Attackers now run on three-hour clocks, while most security programs run on fiscal years. The best attacks don’t look like attacks at all. They arrive as job offers, invoices, calendar invites, and logins that look almost normal. Meanwhile, defenders are being asked to move faster and to tell real from fake at the same time.
When the first GTA VI trailer came out, half of the internet asked the same question. Is this gameplay or real footage? Corporate security teams ask that every day now. In September, a Binance engineer went through two rounds of interviews for what looked like a great job and downloaded the take-home assignment. It was malware. The FBI blamed North Korea, and Binance puts the theft that followed at about $6.3 billion. A day after Microsoft 365 went down this month, attackers were already exploiting the confusion, sending fake outage notices that cited the real incident number.
Now look at how slowly most companies change their defenses. Adopting a new approach means a six-month evaluation, a few more months to roll it out, and a pilot that waits for next year’s plan. I understand the caution, because security leaders have been burned by tools that promised a lot and broke things in production. But an attack that moves at machine speed doesn’t wait for procurement. Attackers are playing on easy mode, and the defense is stuck on a loading screen.
The obvious answer is to make defense faster, and speed does matter. But fast and wrong is still wrong. A team that misreads that signal will shut down its own business over a broken router, while the real attack walks in looking like a job offer. Making the right call takes context that even the best AI model doesn’t have on its own, like who owns the server, whether that user really logged in from Lisbon, and what breaks if a system goes offline. That context lives inside your company, and defense only works when your AI agents know it too.
The leaders I see handling this well have changed three things.
Know the business first. Before buying more detection, they make sure their defenses understand which systems matter, who relies on them, and what normal looks like. AI can’t tell an attack from a failure without that context.
Save the caution for what can’t be undone. Ending one session can be reversed in minutes, so it can happen right away, within rules the team sets. Decisions that can’t be taken back get a closer look and a person’s sign-off.
Change in weeks, not quarters. They try agentic defense early, measure how often it gets real cases right, and give it more room as it earns trust.
In GTA, players can reload their last save. Security teams don’t have that luxury. Companies that make the right call at machine speed will still be around for the next GTA installment, however long it takes.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
This story was originally featured on Fortune.com