GPT-4o — Cyber Threats, Attacks & Incidents

Threat entity extracted from intelligence sources

Frequency
10
occurrences
First Seen
November 5, 2025
Last Seen
August 3, 2026

GPT-4o is a multimodal AI platform that has become a focus in recent cybersecurity reporting as a high-value target for exploitation and safety-evasion.

GPT-4o is a technology platform tracked across 10 threat clusters and 10 intelligence report mentions on ThreatCluster. First observed November 5, 2025; most recent activity August 3, 2026.

Overview

GPT-4o is a multimodal AI platform that has become a focus in recent cybersecurity reporting as a high-value target for exploitation and safety-evasion. Recent disclosures describe seven new vulnerabilities in GPT-4o and its successor GPT-5 that could enable zero-day exploits, alongside guardrail-bypass techniques such as EchoGram tokens, underscoring a substantial attack surface and misuse risk for threat actors.

Related Threat Clusters

  • North Korean BlueNoroff Uses AI for Sophisticated Zoom Phishing Attacks

    A North Korean cyber group, BlueNoroff, has developed a phishing kit that utilizes AI-generated faces to create convincing fake Zoom and Teams meetings targeting cryptocurrency executives. The kit employs pre-edited…

    16 articles · Updated July 24, 2026
  • Deepfake X-Rays Deceive Radiologists and AI Models

    A study published in Radiology reveals that both radiologists and multimodal large language models (LLMs) struggle to distinguish AI-generated deepfake X-ray images from authentic ones. The study involved 17…

    9 articles · Updated March 24, 2026
  • Black Hat USA 2026 Reveals New AI Agent Exploitation Threats

    At Black Hat USA 2026, 29% of sessions focused on AI security, highlighting a shift in attack methodologies targeting agent infrastructure. Significant briefings from Check Point Research revealed vulnerabilities in…

    5 articles · Updated August 2, 2026
  • HackedGPT: Seven New Vulnerabilities in GPT-4o and GPT-5 Discovered

    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…

    2 articles · Updated November 6, 2025
  • CyberSentinel AI Launches Open-Source Cybersecurity Platform with 33 Tools

    CyberSentinel AI has released version 3.0 of its open-source cybersecurity platform, integrating 33 penetration testing and threat intelligence tools. The platform features a provider-agnostic AI engine that supports…

    2 articles · Updated June 20, 2026
  • EchoGram Technique Exploits AI Guardrails Using Specific Prompts

    Security researchers from HiddenLayer have developed a new attack method called EchoGram, which targets the guardrails of large language models (LLMs). By using specific phrases like '=coffee', attackers can bypass…

    2 articles · Updated November 15, 2025
  • Self-Driving Cars and Drones Hijacked by Road Sign Attacks

    Academics have identified a new class of attack known as environmental indirect prompt injection, where self-driving cars and autonomous drones can be manipulated by illicit instructions displayed on road signs. This…

    4 articles · Updated January 30, 2026
  • Chinese Hackers Utilize AI for Cyberattacks on US Companies

    Chinese hackers have leveraged artificial intelligence tools to conduct cyberattacks, with Anthropic, a major US company, reporting a recent incident. The attacks utilize AI models capable of writing code, scanning…

    14 articles · Updated November 29, 2025
  • HackedGPT: Seven Vulnerabilities Found in GPT-4o and GPT-5

    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…

    2 articles · Updated November 6, 2025
  • EchoGram Attack Bypasses AI Guardrails

    Researchers from HiddenLayer have identified a new attack technique called EchoGram that can bypass AI guardrails designed to filter harmful input in large language models (LLMs). This technique allows users to…

    3 articles · Updated November 17, 2025

Recent Intelligence Reports

  • Read the article at Forkast — forkast.news · August 3, 2026
  • BlueNoroff builds Zoom phishing kit with ChatGPT-made faces — Aiweekly.Co · July 25, 2026
  • CyberSentinel AI with 33 Security Tools, Including Nmap, SQLMap, ZAP, and uses Claude, GPT — Cybersecuritynews · June 20, 2026
  • Can you spot a 'deepfake' x-ray? — Auntminnie · March 24, 2026
  • Self-driving cars, drones hijacked by custom road signs — Theregister · January 30, 2026
  • Autonomous cars, drones cheerfully obey prompt injection by road sign — Theregister · January 30, 2026
  • AI 2030: The Coming Era of Autonomous Cyber Crime — Msspalert · November 29, 2025
  • EchoGram tokens like '=coffee' flip AI guardrail verdicts — Theregister · November 15, 2025

Frequently asked questions

What is GPT-4o?

GPT-4o is a multimodal AI platform that has become a focus in recent cybersecurity reporting as a high-value target for exploitation and safety-evasion.

Is GPT-4o still active?

The most recent intelligence report mentioning GPT-4o on ThreatCluster is dated August 3, 2026. Activity was first observed November 5, 2025, giving a tracked span from then to August 3, 2026.

What is GPT-4o associated with?

Across ThreatCluster reporting, GPT-4o most frequently co-occurs with BlueNoroff, Botnet, Data Breach, Malware, Phishing, among 12 tracked related entities.

What are the latest developments involving GPT-4o?

The most significant recent cluster is “North Korean BlueNoroff Uses AI for Sophisticated Zoom Phishing Attacks” (16 articles · Updated July 24, 2026). GPT-4o appears across 10 threat clusters in total, listed above with sources.

How much reporting does ThreatCluster have on GPT-4o?

GPT-4o appears in 10 intelligence report mentions across 10 deduplicated threat clusters, aggregated from 17,000+ monitored sources.

CVSS v3.1 Breakdown