OPENAI Archives - Recon Bee

OpenAI Expands Daybreak With GPT-5.5-Cyber to Help Defenders Patch Security Flaws

23Jun

By RBNo Comments

News

OpenAI Expands Daybreak With GPT-5.5-Cyber to Help Defenders Patch Security Flaws

As part of the Daybreak initiative, which the artificial intelligence (AI) startup announced last month, OpenAI stated on Monday that it is providing trusted defenders with an enhanced version of its GPT-5.5-Cyber model. OpenAI described GPT-5.5-Cyber as its "strongest model yet for finding and helping patch software vulnerabilities," noting that it can "sustain deeper analysis across large codebases" to find security flaws, verify them in a controlled setting, and create and test fixes. In order to prevent new vulnerabilities from entering production codebases and to expedite the process of finding and fixing vulnerabilities in current systems, the tech startup is simultaneously releasing an update to the Codex Security plugin. According to OpenAI, developers can trace attack ve...

OpenAI Launches GPT-5.4-Cyber with Expanded Access for Security Teams

15Apr

By RBNo Comments

News

OpenAI Launches GPT-5.4-Cyber with Expanded Access for Security Teams

Days after rival Anthropic released its own frontier model, Mythos, OpenAI on Tuesday presented GPT-5.4-Cyber, a version of its most recent flagship model, GPT-5.4, specifically tailored for defensive cybersecurity use cases. According to OpenAI, the increasing application of AI enables defenders—those in charge of protecting systems, data, and users—to identify and address issues in the digital infrastructure that everyone depends on more quickly. The artificial intelligence (AI) business announced that it is expanding its Trusted Access for Cyber (TAC) initiative to hundreds of teams in charge of protecting vital software and thousands of authorized individual defenders. AI systems are intrinsically dual-use since hostile actors might utilize technologies created for legitimate...

OpenAI Patches ChatGPT Data Exfiltration Flaw and Codex GitHub Token Vulnerability

31Mar

By RBNo Comments

News

OpenAI Patches ChatGPT Data Exfiltration Flaw and Codex GitHub Token Vulnerability

Check Point has discovered a previously undiscovered vulnerability in OpenAI ChatGPT that allows sensitive conversation data to be exfiltrated without user knowledge or agreement. The cybersecurity business stated in a report released today that a single malicious prompt could transform an otherwise routine discussion into a secret exfiltration channel, revealing user messages, uploaded files, and other sensitive content. The same vulnerability might be exploited by a backdoored GPT to gain access to user data without the user's knowledge or permission. On February 20, 2026, OpenAI addressed the matter after responsible disclosure. There is no proof that the problem was ever used maliciously. Although ChatGPT has a number of safeguards in place to prevent unauthorized data sharin...

Microsoft Detects “SesameOp” Backdoor Using OpenAI’s API as a Stealth Command Channel

4Nov

By RBNo Comments

News

Microsoft Detects “SesameOp” Backdoor Using OpenAI’s API as a Stealth Command Channel

Microsoft has revealed information on a new backdoor called SesameOp that communicates command-and-control (C2) via the OpenAI Assistants Application Programming Interface (API). The Detection and Response Team (DART) at Microsoft Incident Response stated in a technical report released Monday that the threat actor responsible for this backdoor uses OpenAI as a C2 channel to covertly communicate and plan malicious actions within the compromised environment, rather than depending on more conventional techniques. This is accomplished by a backdoor component that uses the OpenAI Assistants API as a relay or storage mechanism to retrieve commands, which the virus subsequently executes. According to the IT giant, the implant was found in July 2025 as part of a complex security event wh...

OpenAI Disrupts Russian, North Korean, and Chinese Hackers Misusing ChatGPT for Cyberattacks

8Oct

By RBNo Comments

News

OpenAI Disrupts Russian, North Korean, and Chinese Hackers Misusing ChatGPT for Cyberattacks

Three activity clusters were terminated by OpenAI on Tuesday for abusing its ChatGPT artificial intelligence (AI) tool to aid in the creation of malware. Among them is a Russian-speaking threat actor who allegedly utilized the chatbot to assist in the creation and improvement of a remote access trojan (RAT), a credential stealer designed to avoid detection. Additionally, the operator prototyped and debugged technical elements that facilitate credential theft and post-exploitation using multiple ChatGPT accounts. We saw that these accounts posted proof of their activity in a Telegram channel devoted to Russian-speaking criminal gangs, suggesting that these accounts are associated with those actors, according to OpenAI. Although the threat actor's direct requests for malicious cont...

Echo Chamber Jailbreak Tricks LLMs Like OpenAI and Google into Generating Harmful Content

24Jun

By RBNo Comments

News

Echo Chamber Jailbreak Tricks LLMs Like OpenAI and Google into Generating Harmful Content

Regardless of the security measures in place, cybersecurity researchers are drawing attention to a novel jailbreaking technique known as Echo Chamber that might be used to fool well-known large language models (LLMs) into producing unwanted results. In contrast to conventional jailbreaks that use character obfuscation or adversarial wording, Echo Chamber uses multi-step inference, semantic steering, and indirect references as weapons. Ahmad Alobaid, a researcher at NeuralTrust, stated in a report that was given to The Hacker News. As a result, the internal state of the model is subtly but effectively manipulated, eventually causing it to generate answers that violate policy. Although LLMs have gradually implemented a number of safeguards to prevent jailbreaks and rapid injections...

OpenAI Bans Accounts Misusing ChatGPT for Surveillance and Influence Campaigns

24Feb

By RBNo Comments

News

OpenAI Bans Accounts Misusing ChatGPT for Surveillance and Influence Campaigns

On Friday, OpenAI announced that it has banned a group of users that developed a suspected surveillance tool driven by artificial intelligence (AI) using its ChatGPT tool. The social media listening tool is reportedly based on one of Meta's Llama models and is most likely Chinese in origin. The accounts in question use the AI company's models to create detailed descriptions and examine documents for a device that can gather data and reports about anti-China protests in the West in real time and share the information with Chinese authorities. The network's actions in endorsing and evaluating surveillance tools have led to the campaign being nicknamed Peer Review read more about OpenAI Bans Accounts Misusing ChatGPT for Surveillance and Influence Campaigns. Get up to date on the la...

ChatGPT-4o can be used for autonomous voice-based scams

4Nov

By RBNo Comments

News

ChatGPT-4o can be used for autonomous voice-based scams

Researchers have demonstrated that financial scams with low to moderate success rates can be carried out by abusing OpenAI's real-time speech API for ChatGPT-4o, an advanced LLM chatbot. OpenAI's most recent AI model, ChatGPT-4o, incorporates text, speech, and vision inputs and outputs, among other improvements. OpenAI has incorporated a number of controls to identify and prevent dangerous content, including the replication of unlicensed voices, as a result of these new features read more about ChatGPT-4o can be used for autonomous voice-based scams. Get up to date on the latest cybersecurity news and enhance your knowledge of cybersecurity with our thorough coverage of the dangers, breaches, and solutions.

OpenAI Blocks 20 Global Malicious Campaigns Using AI for Cybercrime and Disinformation

11Oct

By RBNo Comments

News

OpenAI Blocks 20 Global Malicious Campaigns Using AI for Cybercrime and Disinformation

On Wednesday, OpenAI announced that since the year's beginning, it has thwarted over 20 operations and fraudulent networks that sought to utilize its platform for malevolent ends worldwide. This work included writing biographies for social media accounts, troubleshooting malware, writing articles for websites, and producing AI-generated profile images for fictitious profiles on X. The artificial intelligence (AI) business stated, We have not seen evidence of this leading to meaningful breakthroughs in their ability to create substantially new malware or build viral audiences. Threat actors continue to evolve and experiment with our models read more about OpenAI Blocks 20 Global Malicious Campaigns Using AI for Cybercrime and Disinformation. Get up to date on the latest cybersecur...

Microsoft, OpenAI Warn of Nation-State Hackers Weaponizing AI for Cyber Attacks

15Feb

By RBNo Comments

News

Microsoft, OpenAI Warn of Nation-State Hackers Weaponizing AI for Cyber Attacks

Large language models (LLMs) and artificial intelligence (AI) are being experimented with by nation-state entities connected to China, Iran, North Korea, and Russia to supplement their continuing cyberattack activities. The conclusions stem from a report that Microsoft and OpenAI jointly released, in which they claimed to have terminated the assets and accounts of five state-affiliated actors who had attempted to utilize its AI services for hostile cyber actions. According to a Microsoft report shared with The Hacker News, "Language support is a natural feature of LLMs and is attractive for threat actors with continuous focus on social engineering read more Microsoft OpenAI Warn of Nation-State Hackers Weaponizing AI for Cyber Attacks. Get up to date on the latest cybersecurity n...