Security firm Silent Push used Claude Opus 5 to write exploitation scripts against 12,500 domains, then filtered the results down to several hundred genuinely exploitable dangling DNS records. It’s a known bug class, a DNS record still pointing at a cloud resource that’s since been deleted, letting an attacker reclaim it, but Silent Push demonstrated it against real, named organizations.
You can read more here: Welcome to Danglegeddon – Silent Push
John Watters, Chairman & CEO, iCOUNTER had this to say:
“Silent Push used Claude Opus 5 to write exploitation scripts against 12,500 domains and came back with several hundred workable targets. These are real organizations, not lab conditions: a dangling Azure blob storage record tied to U.S. government infrastructure that could bypass .gov trust filters, an unassigned Société Générale Azure resource, exposed developer credentials and API keys at Ford, and a stale record at Eli Lilly. Silent Push projects losses in the hundreds of billions across pharmaceutical companies alone if this class of vulnerability gets weaponized at scale.
Security teams need to treat domain inventory as something that gets maintained continuously, not set up once and forgotten. That means tracking every DNS record they’ve created, including the orphaned ones pointing at cloud resources that were deleted months or years ago and never cleaned up. Most organizations have no idea how many of those records exist in their own environment, and Silent Push just showed exactly what an attacker can do with the ones they find.
What used to take a nation-state’s intelligence apparatus, mapping thousands of domains, cross-referencing DNS history, and building exploitation infrastructure by hand, now runs as an automated pipeline. An AI model did the reconnaissance, filtered the noise, and handed back a ranked target list touching banking, government, manufacturing, and pharma in a single pass. The skill and headcount required to run a globally coordinated infrastructure attack just dropped by an order of magnitude.”
If you haven’t been attacked by AI, you’re going to be. Of that there is no doubt. The question is will you be ready to defend against an AI attack.
Related
This entry was posted on July 30, 2026 at 2:11 pm and is filed under Commentary with tags Silent Push. You can follow any responses to this entry through the RSS 2.0 feed.
You can leave a response, or trackback from your own site.
AI wrote exploit scripts against 12,500 domains and found live targets
Security firm Silent Push used Claude Opus 5 to write exploitation scripts against 12,500 domains, then filtered the results down to several hundred genuinely exploitable dangling DNS records. It’s a known bug class, a DNS record still pointing at a cloud resource that’s since been deleted, letting an attacker reclaim it, but Silent Push demonstrated it against real, named organizations.
You can read more here: Welcome to Danglegeddon – Silent Push
John Watters, Chairman & CEO, iCOUNTER had this to say:
“Silent Push used Claude Opus 5 to write exploitation scripts against 12,500 domains and came back with several hundred workable targets. These are real organizations, not lab conditions: a dangling Azure blob storage record tied to U.S. government infrastructure that could bypass .gov trust filters, an unassigned Société Générale Azure resource, exposed developer credentials and API keys at Ford, and a stale record at Eli Lilly. Silent Push projects losses in the hundreds of billions across pharmaceutical companies alone if this class of vulnerability gets weaponized at scale.
Security teams need to treat domain inventory as something that gets maintained continuously, not set up once and forgotten. That means tracking every DNS record they’ve created, including the orphaned ones pointing at cloud resources that were deleted months or years ago and never cleaned up. Most organizations have no idea how many of those records exist in their own environment, and Silent Push just showed exactly what an attacker can do with the ones they find.
What used to take a nation-state’s intelligence apparatus, mapping thousands of domains, cross-referencing DNS history, and building exploitation infrastructure by hand, now runs as an automated pipeline. An AI model did the reconnaissance, filtered the noise, and handed back a ranked target list touching banking, government, manufacturing, and pharma in a single pass. The skill and headcount required to run a globally coordinated infrastructure attack just dropped by an order of magnitude.”
If you haven’t been attacked by AI, you’re going to be. Of that there is no doubt. The question is will you be ready to defend against an AI attack.
Share this:
Like this:
Related
This entry was posted on July 30, 2026 at 2:11 pm and is filed under Commentary with tags Silent Push. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.