cyber.hendryadrian.com
Security News · sysdig.com

August 2026 Security Briefing: ChainDrop, AI Agent Risks and Cl0p

· Original article ↗

Summary

Sysdig’s August 2026 briefing covers ChainDrop’s npm poisoning, Ghostjacking risks in AI coding agents, Claude Code use in ransomware intrusions, Cl0p’s exploitation of PTC Windchill, and governance updates. Sysdig says 7 of 8 AI-enabled attacks used ordinary command-ex

Key points

  • ChainDrop, an evolution of the Shai-Hulud npm worm, poisoned more than 400 packages and 2,000 versions in under four hours; its payload used an Ethereum smart contract to resolve command-and-control and targeted AI coding tool credentials.
  • Ghostjacking research showed AI coding agents could be manipulated through actions already allowed by their permissions. The article reports a 90% success rate against Claude Code on Cloudflare’s recommended configuration; a separate Claude Desktop sandbox-escape flaw was patched before disclosure.
  • A Gambit Security report described three unrelated cases of AI use in offensive operations. In one, a suspected The Gentlemen affiliate used Claude Code during intrusions affecting at least six victim organizations.
  • Sysdig’s analysis of eight AI-enabled attacks found that seven used T1059, Command and Scripting Interpreter, and that none used novel techniques.
  • The article reports that Cl0p exploited CVE-2026-12569 in PTC Windchill PDMLink and FlexPLM to compromise around 50 organizations and steal files, documents, blueprints, project files, and databases.
  • The briefing highlights OWASP’s updated Top 10 for LLM Applications, CIRCIA’s pending final rule, and NIST’s request for input on modernizing the National Vulnerability Database.
  • The article recommends auditing trust boundaries between read-only tools and tools that can write or execute, rather than relying only on the permissions granted to AI agents.

Article Details

Event Type
August 2026 cybersecurity briefing covering npm supply-chain poisoning, attacks exploiting AI coding agents, ransomware intrusions involving Claude Code, and Cl0p’s exploitation of PTC Windchill.
Impact
ChainDrop poisoned more than 400 packages and 2,000 versions in under four hours and targeted AI coding tool credentials. A reported Claude Desktop sandbox escape that could remove a control against data exfiltration was patched before its public presentation; a broader flaw pattern affected Cloudflare, Datadog, and Sentry. In one reported ransomware operation, a suspected The Gentlemen affiliate used Claude Code during intrusions affecting at least six organizations. Cl0p reportedly exploited CVE-2026-12569 against around 50 organizations and stole files, documents, blueprints, project files, and databases.

MITRE ATT&CK

T1059 · Command and Scripting InterpreterSysdig reported that seven of eight AI-enabled attacks used command and scripting interpreters.T1078 · Valid AccountsA compromised GitHub maintainer account was used to publish malicious releases with valid SLSA provenance.T1105 · Ingress Tool TransferChainDrop delivered malicious code through poisoned npm packages and releases.T1190 · Exploit Public-Facing ApplicationCl0p exploited CVE-2026-12569 in exposed PTC Windchill systems.

CVE

CVE-2026-12569Cl0p exploited CVE-2026-12569 in PTC Windchill to compromise around 50 organizations and steal sensitive files from major companies.

Threat Actors

Cl0pRansomware group reported to have exploited CVE-2026-12569 and compromised around 50 organizations.The GentlemenA suspected affiliate of this ransomware-as-a-service operation reportedly used Claude Code during intrusions affecting at least six organizations.

Malware

ChainDropAugust’s security briefing covers ChainDrop’s rapid npm supply chain poisoning, Ghostjacking attacks against AI coding agents, and incidents where Claude Code was used in ransomware intrusions.Shai-Hulud 2[Package names / supply chain artifacts ] malicious npm release activity – ChainDrop, Shai-Hulud 2.0, and affected package versions

Vendors

AnthropicAnthropic’s Claude Desktop sandbox escape was identified and patched, while the broader Ghostjacking flaw pattern affected Cloudflare, Datadog, and Sentry.CloudflareAnthropic’s Claude Desktop sandbox escape was identified and patched, while the broader Ghostjacking flaw pattern affected Cloudflare, Datadog, and Sentry.DatadogAnthropic’s Claude Desktop sandbox escape was identified and patched, while the broader Ghostjacking flaw pattern affected Cloudflare, Datadog, and Sentry.PTCIt also highlights Sysdig’s finding that most AI-enabled attacks relied on ordinary command execution, along with Cl0p’s exploitation of PTC Windchill and broader governance updates from OWASP, CIRCIA, and NIST.SentryAnthropic’s Claude Desktop sandbox escape was identified and patched, while the broader Ghostjacking flaw pattern affected Cloudflare, Datadog, and Sentry.

Products

Claude CodeAugust’s security briefing covers ChainDrop’s rapid npm supply chain poisoning, Ghostjacking attacks against AI coding agents, and incidents where Claude Code was used in ransomware intrusions.Claude DesktopAnthropic’s Claude Desktop sandbox escape was identified and patched, while the broader Ghostjacking flaw pattern affected Cloudflare, Datadog, and Sentry.Claude Sonnet 4.6[File / software names ] AI coding tools and agent products targeted or abused – Claude Code, Claude Desktop, Sonnet 4.6FlexPLM[T1190 ] Exploit Public-Facing Application – Cl0p abused CVE-2026-12569 in PTC Windchill PDMLink and FlexPLM to exploit exposed systems and compromise organizations.PTC WindchillIt also highlights Sysdig’s finding that most AI-enabled attacks relied on ordinary command execution, along with Cl0p’s exploitation of PTC Windchill and broader governance updates from OWASP, CIRCIA, and NIST.Windchill PDMLink[T1190 ] Exploit Public-Facing Application – Cl0p abused CVE-2026-12569 in PTC Windchill PDMLink and FlexPLM to exploit exposed systems and compromise organizations.

Related Articles