Is AI moving faster than its safety net?

Is AI moving faster than its safety net? 24 October 2025 by Pieter Arntz argues that artificial intelligence is becoming embedded everywhere, but security and privacy struggles are growing alongside it as safeguards lag behind. The piece notes that the speed of AI development risks outrunning the ability to keep it safe, with AI browsers and the so‑called agentic browser able to read, summarise and act on tasks.

Researchers have demonstrated prompt injection attacks, where an AI browser can follow hidden instructions embedded in images, effectively bypassing user intent. The article also warns of spoofed AI sidebars that mimic genuine interfaces from Atlas and Comet, enabling social engineering at a new level. It cites concerns from industry figures about superintelligence and AI potentially acting beyond human control, underscoring the need to embed basic online safety into AI tools. According to Futurism, these vulnerabilities are being rolled out with deeply worrying security flaws.