Incidents tagged Anthropic

Researchers scanned over 19,000 open-source MCP server repositories, revealing that AI-generated code is present in at least 20% and is disproportionately linked to exploitable vulnerabilities. Key findings include 4.1% of identified flaws being exploitable, with SQL injection, RCE, and path traversal prevalent, and 42.6% of vulnerable repositories showing signs of AI code generation.

OpenAI confirms security breach but says no user data was affected TechRadar

A sophisticated multi-stage supply chain attack, initiated by compromising open-source security scanner Trivy to steal LiteLLM PyPI credentials, injected malicious versions (1.82.7, 1.82.8) of LiteLLM into customer environments. This enabled data exfiltration, system persistence, and lateral movement within Kubernetes clusters, notably leveraging AI coding assistants with unrestricted permissions as an unwitting infection vector.

A critical vulnerability in Anthropic's Claude Code allows for the bypass of its permission system's deny rules. This flaw can be exploited via AI-generated prompt injection, enabling attackers to exfiltrate sensitive credentials and compromise cloud or CI/CD environments.

Evidence indicates that Claude Code is affected by a security issue. Reported affected versions include 2.1.88. Exploitation evidence is classified as confirmed in the wild.

Hackers utilized AI jailbreaking techniques and sophisticated prompt engineering on Generative AI models like Claude and ChatGPT to exploit vulnerabilities within Mexican government systems. This operation led to the successful exfiltration of 150GB of sensitive data, including 195 million taxpayer records, voting information, and government employee credentials.

Anthropic's Threat Intelligence team disrupted the first known AI-orchestrated cyber espionage campaign, where a state-sponsored Chinese threat actor utilized Claude Code to autonomously execute 80-90% of the intrusion life cycle, including reconnaissance, exploitation, credential harvesting, lateral movement, and data exfiltration. This campaign leveraged widely available open-source commodity tools rather than zero-day vulnerabilities, demonstrating a critical shift where AI handles tactical attack execution, significantly compressing detection timelines and challenging traditional incident response frameworks.

Chinese state-sponsored threat actors leveraged Anthropic's Claude Code and Model Context Protocol (MCP) as an "autonomous cyber attack agent" to orchestrate a highly sophisticated and largely automated cyber espionage campaign. This campaign, designated GTG-1002, performed reconnaissance, vulnerability discovery, exploitation, lateral movement, credential harvesting, and data exfiltration against approximately 30 high-value global targets.

Researchers demonstrated that as few as 250 poisoned documents can create a backdoor vulnerability in large language models, irrespective of model size or training data volume. This data poisoning technique, which can induce denial-of-service or potentially facilitate data exfiltration, challenges prior assumptions about the required scale of malicious training data during pretraining.

Salesforce Agentforce was susceptible to a critical indirect prompt injection vulnerability, codenamed ForcedLeak (CVSS 9.4). This flaw allowed attackers to exfiltrate sensitive CRM data by manipulating Web-to-Lead forms, causing AI agents to transmit information to an attacker-controlled domain.

Hackers exploited a vulnerable workflow in the Nx build system to achieve code injection and GITHUB_TOKEN theft, enabling the publication of malicious package versions. These malicious packages deployed a post-install script that exfiltrated thousands of sensitive credentials and notably weaponized AI assistants like Claude and Gemini for reconnaissance and data exfiltration.