Pull down to refresh stories
Emerging

Top AI Agents Built to Catch Malicious Code Can Be Tricked Into Running It

Ask an AI coding agent to scan open-source code for security holes, and it might run the attacker's code on your own machine instead. That is the finding in a proof-of-concept published Wednesday by the AI Now Institute, an attack it calls " Friendly Fire. This piece sits on 1 source layers, but the real value is showing why the story should not be skimmed past too quickly.

Ask an AI coding agent to scan open-source code for security holes, and it might run the attacker's code on your own machine instead. That is the finding in a proof-of-concept published Wednesday by the AI Now Institute, an attack it calls " Friendly Fire. The signal is strong enough to deserve attention, but it still needs to be read as something developing rather than fully settled.

Emerging The topic has initial corroboration, but the newsroom is still waiting on stronger confirmation.
Reference image for: Top AI Agents Built to Catch Malicious Code Can Be Tricked Into Running It
Reference image from The Hacker News. The Hacker News

Ask an AI coding agent to scan open-source code for security holes, and it might run the attacker's code on your own machine instead. That is the finding in a proof-of-concept published Wednesday by the AI Now Institute, an attack it calls " Friendly Fire. " It works against Anthropic's Claude Code and OpenAI's Codex when either is running in an autonomous mode that approves its own commands. The Hacker News is the main source layer for now, and the rest should be read as a signal that is still widening. In security, the real value is not just the warning itself but the way it changes operational risk, account safety, and the cost of responding later.

What is happening now

Ask an AI coding agent to scan open-source code for security holes, and it might run the attacker's code on your own machine instead. The Hacker News form the main source layer behind the core facts in this piece. This is still a developing thread, so the useful part is knowing which source signals are hardening and which ones still need caution. In security, the real value is whether the team becomes measurably safer, not whether another settings screen has been added.

Where the sources line up

The Hacker News is the main source layer for now, and the rest should be read as a signal that is still widening. That is the finding in a proof-of-concept published Wednesday by the AI Now Institute, an attack it calls " Friendly Fire. The Hacker News form the main source layer behind the core facts in this piece. In security, the real value is whether the team becomes measurably safer, not whether another settings screen has been added. The people who should read carefully are system admins, shop owners, content teams, and anyone holding customer data or operational accounts.

The details worth keeping

" It works against Anthropic's Claude Code and OpenAI's Codex when either is running in an autonomous mode that approves its own commands. In security, the real value is not just the warning itself but the way it changes operational risk, account safety, and the cost of responding later. The people who should read carefully are system admins, shop owners, content teams, and anyone holding customer data or operational accounts. The next step is to see whether the current signals harden into a durable change or fade as a short-lived experiment.

Why this matters most

The signal is strong enough to deserve attention, but it still needs to be read as something developing rather than fully settled. With 1 source layers on the table, the part worth reading most closely is where firm facts meet the market's early reaction. It hijacks the exact job these tools are sold for: checking untrusted third-party code for problems. The next step is to see whether the current signals harden into a durable change or fade as a short-lived experiment. That is why the useful reading move is not to stop at the headline, but to compare the promise, the workflow change, and the likely cost before deciding anything.

What to watch next

The next layer to watch is scope, patch speed, and the operating cost if teams are forced to change process because of this story. Patrick Tech Media will keep checking rollout speed, user reaction, and how The Hacker News update the next pieces. From 1 early signals, the piece keeps 1 references that are useful for locking the main details in place. That is why the useful reading move is not to stop at the headline, but to compare the promise, the workflow change, and the likely cost before deciding anything.

Source notes