Azeem Azhar warns AI is speeding up the discovery of software vulnerabilities
In conversation with Paul Krugman, researcher Azeem Azhar argued that AI models can now find software flaws at scale and without tiring, unlike human security teams. He cited the PHP hack, in a language he said runs 70% of websites, and called it "the Wuhan moment." He noted that capabilities of closed models such as Claude Mythos spread quickly to open-weight models.
In 35 seconds
- Azhar describes a new regime in which AI relentlessly hunts for vulnerabilities and chains them together
- The PHP hack, used by 70% of websites according to Azhar, led him to speak of a "Wuhan moment"
- Capabilities of closed models such as Claude Mythos reached open-weight models this summer
Read the full story at Paul Krugman | Substack
IASummary generated by artificial intelligence from a story by Paul Krugman | Substack. It may contain errors: the original source prevails.
Comments
Be respectful. Comments reported by several people are hidden until a moderator reviews them.
You do not need an account to read, only to comment. Sign in with a code sent to your email or with your Google account. We only show the name you choose, never your email.