OpenAI Plans to Launch a New Cybersecurity-Focused GPT-6 Series AI Model: Report

OpenAI, the Sam Altman-led AI giant, released two new GPT-6 series models Wednesday, namely the GPT-6 Sol and GPT-6 Luna models, while GPT-6 Astra was launched earlier this month. Recently, OpenAI has launched multiple use-case-specific AI models, serving particular needs, including legal and financial services, namely Astra for Law and ChatGPT for Financial Services, both powered by the GPT-6 series models. Now, the company is reportedly preparing to release another model, specifically built for cybersecurity. Expected to be called GPT-6 Cyber, the AI model is said to debut in preview soon and is said to be deployed via a new product.

OpenAI GPT-6 Cyber: Here's What We Know So Far

Citing “multiple” people familiar with the matter, Fortune reports that OpenAI is planning to launch a new cybersecurity-focused AI model soon, namely GPT-6 Cyber. The AI model will reportedly first be released in preview in the coming weeks. However, the report does not specify the exact date. The new GPT-6 Cyber AI model will reportedly be accompanied by a new consumer “product”, which will let customers deploy it in a “more secure” manner.

The new “product” will also help in automating the deployment of the GPT-6 Cyber AI model, the report suggests. Further, the report highlights that OpenAI has yet to name the product. Moreover, it is unclear whether the Sam Altman-led AI giant is planning to launch hardware or a device for the AI model's deployment, or a digital service. It is said to help users automate their workflows.

As per the report, the GPT-6 Cyber AI model is already in the alpha testing phase. It is said to be available to a limited number of users via OpenAI's “application-only” Daybreak Red programme, which is divided into red and blue access tiers. The Blue tier is said to offer general access to the models, while the Red tier reportedly provides access to the AI giant's most advanced “cybersecurity products”.

The model is reportedly capable of automatically patching cybersecurity vulnerabilities, stopping bad actors from exploiting vulnerabilities with the help of AI-powered tools. On top of this, the new cybersecurity-focused AI model will reportedly allow OpenAI to monitor how its models are being utilised.

Interestingly, the development comes after details regarding two major cybersecurity incidents involving OpenAI surfaced online. First, an internal OpenAI model broke out of its sandbox environment, autonomously gaining access to the public internet and orchestrating an attack against another AI firm, Hugging Face. Later, OpenAI itself was targeted by a group of ethical hackers, who gained access to the company's internal repositories and other connectors by exploiting two critical vulnerabilities, which were later patched by OpenAI.