AI used for cyberattacks, election interference, and weapon development

Artificial intelligence is increasingly being utilized for sophisticated cyberattacks, election interference, and weapon development. In Sweden, the government is actively monitoring hybrid threats, with the Swedish Psychological Defence Agency identifying networks of approximately 1,200 fake accounts used for information manipulation.

Cybersecurity researchers have observed a shift toward 'agentic AI' that can automate complex attack chains. A recent mass campaign compromised at least 440 systems across 48 countries, including entities in Poland. This campaign utilized hundreds of AI agents—running on the DeepSeek model within an OpenAI Codex environment—to develop exploits, test attacks, and manage post-access activities.

Furthermore, Anthropic reported that a group in northern Yemen attempted to use its Claude model to assist in developing guided missile technology. The AI was used to write flight control, navigation, and stabilization software, acting as a virtual engineering team. While a physical test of the missile was conducted, it reportedly failed to function correctly.