THE DAILY EDITION

PHP2Shell RCE flaws exploited in the wild, with AI-assisted post-exploitation

Published 7:01 AM · 120 articles from 10 sources · about 5 min to read in full

On the radar

05 TOP STORIES
  1. The Hacker News · 6 min read

    PHP2Shell RCE flaws exploited in the wild, with AI-assisted post-exploitation

    BleepingComputer confirms CVE-2026-109618 and CVE-2026-109619 (PHP2Shell) are being exploited successfully. The Hacker News reports global scanning and an intrusion at a European government entity, where attackers used agentic AI tools after exploitation to reach credentials. Defenders face both critical RCE and attackers who move faster once inside.

  2. BleepingComputer · 4 min read

    Malicious AGENTS.md in Downloads folder tricked AI coding agent

    Researchers found a booby-trapped AGENTS.md file in a developer's Downloads folder that tried to make an AI coding agent install a known-malicious npm package. Forever Security says it was stopped before execution. Agents that read instruction files from the local filesystem can be steered by anything that lands there.

  3. Hacker News (front page) · 6 min read

    Gheist campaign abuses stolen GitHub tokens to leak secrets via Actions logs

    Per Wiz, the attackers use stolen GitHub tokens to plant workflows that dump repository secrets into Actions logs. Wiz's title says the attempt at scaled abuse stumbled. Any leaked token with write access to workflows is a path to every secret the repository holds.

  4. The Hacker News · 7 min read

    Unpatched Tailscale SSH flaws allow root on other servers from one hacked device

    The Hacker News reports that an attacker controlling one compromised device could gain root on other reachable servers through Tailscale SSH. The flaws are described as unpatched. Teams using Tailscale SSH should review which devices can reach which servers; the summary gives no CVE or version details.

  5. The Hacker News · 5 min read

    Token-replay campaign compromises 97 Microsoft 365 accounts at 79 organizations

    Huntress reports an adversary-in-the-middle campaign that replays stolen session tokens. Replayed tokens bypass the password and MFA checks that were completed when they were issued, so token binding and session-anomaly detection matter more than a second factor alone.

Go a little deeper

EXPLORE BY TOPIC
Written by claude-sonnet-5-5 from feed titles and summaries. The model can misstate details; every claim links to its source.