
Google finds attackers automating hacks with no human involved
A framework called exploit_agent compared software before and after patches, wrote exploits and ran them on its own, Google’s threat researchers said.

A framework called exploit_agent compared software before and after patches, wrote exploits and ran them on its own, Google’s threat researchers said.

Security firm Claroc found Grok Build's share links could be enumerated, exposing business plans, medical results and confidential SpaceXSI sales files.

Prime Minister Anthony Albanese accepted all 18 recommendations of a review into an OpenAI agent's June access to a Medicare data portal.

Anthropic’s first frontier models since Dario Amodei’s September call to slow AI development cost 20% less than the versions they replace.

Prosecutors say the men used AI to alter voices and images of executives and lawyers, tricking staff at hospitals, a bank and a mutual fund into wiring money.

The automated system also validated the bugs and wrote patches for flaws that let a local user seize root control of a Linux machine, OpenAI said.

Parameter Labs showed a crafted public share could run code in a victim's signed-in session and read private chats before OpenAI patched it October 26.

Australia's signals agency responded to 1,500 incidents in 2025-26, up 20%, and said services may go offline while they are secured.

Google says Gemini 4 Carbon leads tests of finding, exploiting, and patching software flaws, and wider access will follow as it refines guardrails.