OpenAI launches model capable of cyberattacks but with system that limits actionsPhoto by idilioarte on Pexels
AI News
Искусственный интеллект
Инвестиции
Santarém

OpenAI launches model capable of cyberattacks but with system that limits actions

TV RECORD EUROPA6 September 2026 at 12:00

OpenAI announced the launch of the Astra model, which is capable of invading protected systems autonomously. This model represents an advanced cybersecurity capability that can be used for testing and identifying vulnerabilities.

The Astra model will include an integrated monitoring system, designed to track and stop the system's actions when they exceed defined limits. This security feature aims to ensure that the model does not perform unauthorized operations.

The company developed this model in the context of the growing debate about artificial intelligence and cybersecurity. The ability to perform autonomous cyberattacks raises questions about the risks associated with this type of technology.

This news was shared on the TV RECORD website, which regularly presents updates about technology and innovation.

Related articles

O teu chefe está de olho em ti
AI News

Your boss is watching you

With the advancement of artificial intelligence, cameras, sensors and monitoring programs, companies today have more tools than ever to monitor their employees in real time. The article explores how this workplace surveillance trend has intensified, mentioning the example of Burger King, which launched the "Patty" AI chatbot, an assistant that monitors whether cashiers say "please" and "thank you" and alerts when soda machines need restocking, questioning whether this type of control actually produces the desired results for companies.

ZAP Notícias06/09/26, 11:10
Agentes de IA: Longe da magia, de olhos postos no futuro
AI News

AI Agents: Far from magic, with eyes on the future

This opinion article by Tiago Costa Lima, Director of Product Strategy at Cegid Portugal, addresses the growing impact of AI agents and automation in the business environment. The author argues that, although intelligent agents represent a significant technological evolution, it is essential to maintain realistic expectations about their capabilities and limitations. The article explores how companies can integrate these tools strategically, preparing for a future where collaboration between humans and machines will be increasingly decisive for organizational competitiveness.

SAPO Notícias06/09/26, 11:01
A inteligência artificial entrou nas escolas, nas universidades, no trabalho e n…
AI News

Artificial intelligence has entered schools, universities, work and…

Artificial intelligence has become present in schools, universities, workplaces and people's daily lives, offering extraordinary tools, but also raising fundamental questions about the future of learning and creativity. The article explores the challenges of teaching and learning in a world where automatic answers are available at any moment, questioning how to stimulate genuine creativity when there is the temptation to skip stages of the creative process and what role human beings continue to have in this new technological era.

SAPO Notícias06/09/26, 09:00
AI News

In dispute with NYT, OpenAI says copyright cannot stand in the way of progress

OpenAI and Microsoft presented their arguments in court against the lawsuit filed by The New York Times for copyright infringement, with OpenAI arguing that copyright cannot impede technological progress. The company used references to poetry and visual arts in its defense. The trial was temporarily avoided with summary judgment petitions filed at the U.S. District Court for the Southern District of New York, and Judge Sidney H. Stein is expected to decide in the coming weeks. Other newspapers, such as The Seattle Times and Newsday, also filed similar actions against OpenAI and Microsoft for the same reason: the unauthorized use of content for AI training.

oglobo06/09/26, 08:00