Gpt-5 is a tool tracked across 9 threat clusters and 8 intelligence report mentions on ThreatCluster. First observed November 5, 2025; most recent activity February 18, 2026.
Researchers from Netskope Threat Labs tested the capabilities of LLMs, specifically GPT-3.5-Turbo and GPT-4, to generate malicious code. Although they succeeded in producing malware, the output was deemed too unreliable…
Seven new vulnerabilities have been identified in the GPT-4o and GPT-5 models, potentially allowing for zero-day exploits. These vulnerabilities affect users of the models and could lead to unauthorized access or…
Researchers have identified multiple vulnerabilities in OpenAI's ChatGPT, including seven critical flaws that allow attackers to exfiltrate personal data, inject malicious prompts, and manipulate the AI's memory. The…
Anthropic identified a vulnerability in blockchain smart contracts using its Claude AI models, which could have led to a potential $4.6 million exploit. However, the company chose not to pursue the attack, as it would…
On March 6, 2026, OpenAI announced the launch of Codex Security, a sophisticated application security tool designed to identify software vulnerabilities, available for select ChatGPT users. Concurrently, OpenAnt, an…
Researchers at Netskope Threat Labs explored the capabilities of LLMs in generating malware. They successfully prompted both GPT-3.5-Turbo and GPT-4 to create malicious code, but the results were deemed too unreliable…
Researchers in Italy published a paper revealing that poetic prompts can effectively jailbreak large language models (LLMs). This technique, termed 'adversarial poetry,' demonstrated high attack-success rates across…
Tenable Research identified seven critical vulnerabilities in OpenAI's ChatGPT models, including ChatGPT-4o and ChatGPT-5, that could allow attackers to steal personal data and hijack user conversations. The…
Seven new vulnerabilities have been identified in the AI models GPT-4o and GPT-5, potentially allowing for zero-day exploits. These vulnerabilities could impact various applications utilizing these models, raising…