“It’s one in a vast spectrum of opinions.”

In the wake of recent events regarding Artificial Intelligence (AI), Arlindo Oliveira, researcher at INESC‑ID and President of INESC, was invited to RTP to clarify what risks AI truly presents.

This intervention followed the resignation of Anthropic researcher Jacob Coxon, who argued that advanced AI systems could pose a threat to humanity before the end of the decade. Arlindo Oliveira considers this to be “one in a vast spectrum of opinions” and explained that recent cases of AI agents accessing external machines occurred in controlled test environments designed to identify software vulnerabilities. Furthermore, the researcher noted that going from these instances to a hacking operation or human extinction is a considerable leap.

Still, Arlindo Oliveira acknowledged that the scenario is not conceptually impossible nor entirely risk‑free, giving examples such as cybersecurity and access to critical water or energy systems. However, he stresses that these scenarios are difficult to accomplish, given the protection of AI systems, and highlights the importance of regulation capable of keeping up with rapid advances in AI.

Full article on the RTP Notícias website.

© 2026 INESC-ID. When sharing this article, credit INESC-ID with a link to the original source.

Image: © RTP