OpenAI, Anthropic, Google API Flaw Let Weaker AI Models Decode Stronger Models' Reasoning

A newly disclosed flaw in the way OpenAI, Anthropic, and Google carried hidden AI reasoning between API calls let researchers recover internal reasoning and secrets from session logs, including API keys and passwords. The weakness affected encrypted reasoning objects used by the providers'...

Prompt Injections for Defense

This seems to work: Researchers from Tracebit on Monday said they found that placing prompt injections alongside passwords, cryptographic keys, and other secrets stored on Amazon Web Services was often all that was needed to shut down attacks from AI hacking agents. The prompts direct the attacki...

ToolHazard: Scaling Adversarial Environments for Security Evaluation and Alignment of LLM-Based Agents

Large language model LLM agents integrated with external tools are vulnerable to indirect prompt injections embedded in environmental states. However, existing studies largely rely on manually implemented or reused environments, stochastic LLM-based tool simulation, and predefined injection...

Rethinking Agent Security As a Networking Problem

AI agents are rapidly becoming more capable and widely deployed, promising substantial gains in productivity and enabling new classes of applications. However, their growing autonomy also introduces significant privacy and security risks. Existing defenses are predominantly agent-centric, relying...

CVE-2026-73032

PapersGPT for Zotero 0.6.1 contains a remote code execution vulnerability that allows attackers to execute arbitrary JavaScript by returning malicious code from an LLM endpoint that is passed unsanitized to window.eval in views.ts. Attackers can exploit this through prompt injection in PDFs, MITM...

CVE-2026-73036

Bash-it 3.2.0 contains a terminal escape sequence injection vulnerability in the barbuk theme's Python virtualenv prompt segment that allows local attackers to inject arbitrary terminal control sequences by embedding escape sequences in the requires-python field of a pyproject.toml file. When a...

CVE-2026-73036

Bash-it 3.2.0 contains a terminal escape sequence injection vulnerability in the barbuk theme's Python virtualenv prompt segment that allows local attackers to inject arbitrary terminal control sequences by embedding escape sequences in the requires-python field of a pyproject.toml file. When a...

EUVD-2026-56927

Bash-it 3.2.0 contains a terminal escape sequence injection vulnerability in the barbuk theme's Python virtualenv prompt segment that allows local attackers to inject arbitrary terminal control sequences by embedding escape sequences in the requires-python field of a pyproject.toml file. When a...

EUVD-2026-56896

PapersGPT for Zotero 0.6.1 contains a remote code execution vulnerability that allows attackers to execute arbitrary JavaScript by returning malicious code from an LLM endpoint that is passed unsanitized to window.eval in views.ts. Attackers can exploit this through prompt injection in PDFs, MITM...

CVE-2026-73032

PapersGPT for Zotero 0.6.1 contains a remote code execution vulnerability: an unsanitized LLM response passed to window.eval() in views.ts allows arbitrary JavaScript execution in Zotero's chrome-privileged context. Attackers can exploit prompt injection in PDFs, MITM interception of API requests...

CVE-2026-73032 PapersGPT for Zotero 0.6.1 RCE via Unsanitized LLM Response eval()

PapersGPT for Zotero 0.6.1 contains a remote code execution vulnerability that allows attackers to execute arbitrary JavaScript by returning malicious code from an LLM endpoint that is passed unsanitized to window.eval in views.ts. Attackers can exploit this through prompt injection in PDFs, MITM...

CVE-2026-73032 PapersGPT for Zotero 0.6.1 RCE via Unsanitized LLM Response eval()

PapersGPT for Zotero 0.6.1 contains a remote code execution vulnerability that allows attackers to execute arbitrary JavaScript by returning malicious code from an LLM endpoint that is passed unsanitized to window.eval in views.ts. Attackers can exploit this through prompt injection in PDFs, MITM...

CVE-2026-73032 PapersGPT for Zotero 0.6.1 RCE via Unsanitized LLM Response eval()

PapersGPT for Zotero 0.6.1 contains a remote code execution vulnerability that allows attackers to execute arbitrary JavaScript by returning malicious code from an LLM endpoint that is passed unsanitized to window.eval in views.ts. Attackers can exploit this through prompt injection in PDFs, MITM...

On Understanding, Identifying, and Mitigating Vulnerabilities in Agentic Large Language Models

Large Language Models LLMs have undergone a shift from stateless conversational interfaces to autonomous agents capable of multi-step planning, tool invocation, code execution, and maintaining persistent memory. When these agents operate with real-world privileges---calling APIs, modifying files,...

A Study of Cursorrules Files in GitHub Open Source Projects

Prompts are the primary mechanism for communicating with AI agents, and they directly influence the quality and reliability of AI-generated code. As AI-assisted programming becomes widely adopted, modern tools increasingly combine dynamic conversational prompts with static configuration-like prom...

PT-2026-70891

Bash-it 3.2.0 contains a terminal escape sequence injection vulnerability in the barbuk theme's Python virtualenv prompt segment that allows local attackers to inject arbitrary terminal control sequences by embedding escape sequences in the requires-python field of a pyproject.toml file. When a...

PraisonAI CodeAgent 1.6.77 Remote Code Execution

The CodeAgent.executepython in PraisonAI executes LLM-generated code without sandboxing, AST validation, or import restrictions. This allows full environment secret exfiltration and arbitrary code execution via prompt injection. Versions 1.6.77 and below are affected. !/usr/bin/env python3 Exploi...

CVE-2026-73032: Improper Control of Generation of Code

PapersGPT for Zotero 0.6.1 contains a remote code execution vulnerability that allows attackers to execute arbitrary JavaScript by returning malicious code from an LLM endpoint that is passed unsanitized to window.eval in views.ts. Attackers can exploit this through prompt injection in PDFs, MITM...

PT-2026-70874

Name of the Vulnerable Software and Affected Versions PapersGPT for Zotero version 0.6.1 Description A remote code execution issue exists where arbitrary JavaScript can be executed by returning malicious code from an LLM endpoint. This occurs because the response is passed unsanitized to the...

PraisonAI praisonaiagents 1.6.77 - Remote Code Execution

!/usr/bin/env python3 Exploit Title: PraisonAI CodeAgent 1.6.77 - Remote Code Execution CVE: CVE-2026-61447 Date: 2026-07-13 Exploit Author: Mohammed Idrees Banyamer Author Country: Jordan Instagram: @banyamersecurity Author GitHub: https://github.com/mbanyamer Author Blog :...