OpenAI warns of persistent AI cyber-attacks, pauses advanced model development

OpenAI’s chief global affairs officer, Chris Lehane, told *The Guardian* that the industry must adopt stricter safety standards as critics accuse AI firms of acting “recklessly.” In a statement released this week, Lehane warned that as artificial‑intelligence models acquire the ability to plan and execute sophisticated cyber‑attacks, society must prepare for “ongoing, persistent” threats. The company has already paused the development of its most advanced internal models, citing rising safety concerns and the need to reassess the potential risks of its technology.

Lehane described the current era as a “different chapter” in AI, noting that the capabilities of the technology have outpaced existing safeguards. He urged regulators, industry stakeholders, and the public to collaborate on new protocols that would prevent malicious use while still allowing beneficial innovation. The pause reflects a broader trend in the sector, where several firms are re‑examining their research agendas in light of the growing power of large language models and the possibility of autonomous offensive operations.

In conclusion, OpenAI’s announcement signals a shift toward heightened caution in AI development. By calling for updated safety measures and acknowledging the potential for AI‑driven cyber‑attacks, the company highlights the urgent need for coordinated action to ensure that the technology’s benefits do not come at the expense of security.