GitLab Duo Vulnerability Enabled Attackers to Hijack AI Responses with Hidden Prompts

Cybersecurity researchers have discovered an indirect prompt injection flaw in GitLab's artificial intelligence AI assistant Duo that could have allowed attackers to steal source code and inject untrusted HTML into its responses, which could then be used to direct victims to malicious websites...

CVE-2023-32786

In Langchain through 0.0.155, prompt injection allows an attacker to force the service to retrieve data from an arbitrary URL, essentially providing SSRF and potentially injecting content into downstream tasks...

CVE-2023-32391

The issue was addressed with improved checks. This issue is fixed in iOS 15.7.6 and iPadOS 15.7.6, watchOS 9.5, iOS 16.5 and iPadOS 16.5, macOS Ventura 13.4. A shortcut may be able to use sensitive data with certain actions without prompting the user...

CVE-2023-24579

McAfee Total Protection prior to 16.0.51 allows attackers to trick a victim into uninstalling the application via the command prompt...

CVE-2023-23774

Motorola EBTS/MBTS Site Controller drops to debug prompt on unhandled exception. The Motorola MBTS Site Controller exposes a debug prompt on the device's serial port in case of an unhandled exception. This allows an attacker with physical access that is able to trigger such an exception to extrac...

CVE-2023-27963

The issue was addressed with additional permissions checks. This issue is fixed in macOS Ventura 13.3, iOS 16.4 and iPadOS 16.4, iOS 15.7.4 and iPadOS 15.7.4, macOS Monterey 12.6.4, tvOS 16.4, watchOS 9.4. A shortcut may be able to use sensitive data with certain actions without prompting the use...

CVE-2023-21383

In Settings, there is a possible way for the user to unintentionally send extra data due to an unclear prompt. This could lead to local information disclosure with no additional execution privileges needed. User interaction is needed for exploitation...

CVE-2023-23493

A logic issue was addressed with improved state management. This issue is fixed in macOS Ventura 13.2, macOS Monterey 12.6.3. An encrypted volume may be unmounted and remounted by a different user without prompting for the password...

CVE-2023-34641

KioWare for Windows through v8.33 was discovered to contain an incomplete blacklist filter for blocked dialog boxes on Windows 10. This issue can allow attackers to open a file dialog box via the function window.print which can then be used to open an unprivileged command prompt...

CVE-2022-34474

Even when an iframe was sandboxed with allow-top-navigation-by-user-activation, if it received a redirect header to an external protocol the browser would process the redirect and prompt the user as appropriate. This vulnerability affects Firefox 102...

CVE-2022-3750

The has a CSRF vulnerability that allows the deletion of a post without using a nonce or prompting for confirmation...

A Critical Evaluation of Defenses against Prompt Injection Attacks

Large Language Models LLMs are vulnerable to prompt injection attacks, and several defenses have recently been proposed, often claiming to mitigate these attacks successfully. However, we argue that existing studies lack a principled approach to evaluating these defenses. In this paper, we argue...

FreeFloat FTP Server Buffer Overflow Vulnerability (CNVD-2025-14495)

FreeFloat FTP Server is an FTP service from FreeFloat. A buffer overflow vulnerability exists in FreeFloat FTP Server due to an unknown function in the component PROMPT command handler. No detailed vulnerability details are provided at this time...

CVE-2022-34102

Insufficient access control vulnerability was discovered in the Crestron AirMedia Windows Application, version 4.3.1.39, in which a user can pause the uninstallation of an executable to gain a SYSTEM level command prompt...

CVE-2021-42061

SAP BusinessObjects Business Intelligence Platform Web Intelligence - version 420, does not sufficiently encode user-controlled inputs, resulting in Cross-Site Scripting XSS vulnerability. This allows a low privileged attacker to retrieve some data from the victim but will never be able to modify...

CVE-2020-12798

Cellebrite UFED 5.0 to 7.5.0.845 implements local operating system policies that can be circumvented to obtain a command prompt via the Windows file dialog that is reachable via the Certificate-Based Authentication option of the Wireless Network Connection screen...

CVE-2020-15682

When a link to an external protocol was clicked, a prompt was presented that allowed the user to choose what application to open it in. An attacker could induce that prompt to be associated with an origin they didn't control, resulting in a spoofing attack. This was fixed by changing external...

CVE-2025-5074 FreeFloat FTP Server PROMPT Command buffer overflow

A vulnerability, which was classified as critical, was found in FreeFloat FTP Server 1.0. Affected is an unknown function of the component PROMPT Command Handler. The manipulation leads to buffer overflow. It is possible to launch the attack remotely. The exploit has been disclosed to the public...

CVE-2025-5074 FreeFloat FTP Server PROMPT Command buffer overflow

A vulnerability, which was classified as critical, was found in FreeFloat FTP Server 1.0. Affected is an unknown function of the component PROMPT Command Handler. The manipulation leads to buffer overflow. It is possible to launch the attack remotely. The exploit has been disclosed to the public...

CVE-2025-5074

CVE-2025-5074 affects FreeFloat FTP Server 1.0; buffer overflow in the PROMPT Command Handler can be exploited remotely. Public exploit signals risk. No patch/version fix is disclosed in the provided docs; PT-2025-22461 suggests temporary workarounds: disable the PROMPT Command Handler and restri...