Re:Project Zero will disclose vulnerabilities 90 days after ...
by Gigachad on Saturday November 14, 2026 @11:03PM
(#50503903)
Because LLMs are not deterministic or predictable and make mistakes that they are capable of fixing.
LLMs will write vulnerable code while the same model can find those vulnerabilities if you prompt it right. Some stats from a previous post was that the CVEs found in Linux were costing on average $1000 in api usage each. So having the LLM prevent writing vulnerabilities would explode the costs.
LLMs will write vulnerable code while the same model can find those vulnerabilities if you prompt it right. Some stats from a previous post was that the CVEs found in Linux were costing on average $1000 in api usage each. So having the LLM prevent writing vulnerabilities would explode the costs.