The Safe Internet Is Shrinking - The Atlantic
Open-source AI models can now find software vulnerabilities and automate attacks at almost no cost, and the article argues that malicious cyber activity is already spiking as a result. It expects more defaced websites, more stolen personal data, and more ransomware.
- Open-source models, which are free to download and use without restrictions, have recently become very good at finding vulnerabilities and automating attacks.
- Malicious cyber activity has already spiked in the past couple of months.
- The expected results are more website defacements, more personal information theft, and wider ransomware spread.
Here's the core claim: AI agents have made cyberattacks cheap and scalable in a way they weren't before. The excerpt quotes Rubin saying that open-source models, the kind anyone can download and run without restrictions, have in just the past couple of months become very good at finding software vulnerabilities and automating attacks. The result, in his telling, is a spike in malicious activity. Expect more defaced websites, more stolen personal data, and more ransomware. The shorter version is that the cost of attacking has dropped sharply while the pool of people able to attack has grown.
The argument rests on a simple economic idea. Hacking used to require skilled people, and skilled people are scarce and expensive. An AI agent can run around the clock, try thousands of approaches in parallel, and never tire. That turns attacks into something closer to a utility you can buy by the hour. The worry isn't just that sophisticated groups get better tools. It's that unsophisticated actors, who previously couldn't have executed a serious intrusion, can now point an automated system at a target and let it work. Because open-source models carry no usage restrictions, there's no single company to lock them down, which is what makes the problem hard to contain through the usual channels.
The title's framing, that the safe internet is shrinking, points to what this means for ordinary users and organizations. The sharpest pressure falls on small sites, under-resourced businesses, and individuals who lack dedicated security teams, since they're the easiest targets for automated scanning. The piece implies a world where basic defenses like patching quickly, limiting exposed data, and maintaining good backups stop being optional hygiene and become the price of staying online at all. For a reader deciding whether this affects them, the practical takeaway is that the baseline threat level has moved, and assumptions about what counts as "obscure enough to be safe" no longer hold.