OpenAI Launches GPT-6 Model with Critical Cybersecurity Capabilities for the First Time

The American company OpenAI has launched its latest language model, GPT-6, codenamed Astra, in a move the company described as the release of its most advanced model to date.

Internal evaluations conducted by OpenAI revealed that the model reached a critical level in cybersecurity metrics, as it is able to discover security vulnerabilities and exploit fortified systems with limited human guidance.

The company announced the activation of enhanced safety protocols to manage the model, responding to pressure from regulators and researchers demanding greater transparency in disclosing the capabilities of powerful models.

The launch represents the latest milestone in the race for large language models among tech giants, amid growing calls to establish clear regulatory frameworks to curb harmful uses of ultra-capable artificial intelligence.

What do these terms mean?

Security Vulnerability: A flaw in a computer system that an attacker can exploit to gain unauthorized access or control over systems, similar to a broken door in a house.

Astra: The code name for the GPT-6 model, referring to a general-purpose, multi-capable model operating in diverse environments.

Cybersecurity: A field focused on protecting systems, networks, and data from digital attacks. It includes tools and policies that prevent data theft or service disruption.