Skip to content
tag — vulnerability-research

grep -rl "vulnerability-research" ./articles

#vulnerability-research

12 articles

It found a 27-year-old bug for under twenty thousand dollars

2026-11-12AI

Anthropic published what its gated security model cost to run. A thousand scanning runs against OpenBSD came to under twenty thousand dollars. A working Linux privilege escalation came to under a thousand. Over 99 percent of what it found is unpatched.

OpenAI says 81 Firefox bugs, Mozilla's advisory credits it with 41

2026-11-07AI

OpenAI launched GPT-6 Cyber with a claim of 81 previously unknown high-severity Firefox vulnerabilities. We counted the OpenAI credits in the Mozilla advisory where those fixes landed. There are 41, and the advisory never says a model found any of them.

Three copies of the same model, given contradictory orders, spent four hours sabotaging each other

2026-08-19AI

Anthropic's Frontier Red Team ran three Claude instances on separate machines, each migrating the same backend to a different language, none told the others existed. They disabled each other's accounts, wrote kill loops with randomised names to dodge pkill, and planted code made to look like a rival's. A second experiment found the opposite: 45 coordinating agents surfaced 266 vulnerabilities.