New Delhi: The Ex-Google CEO Eric Schmidt, who led Google from 2001 to 2011, has delivered a dangerous assessment of the upcoming artificial intelligence security risks while simultaneously expressing optimism about the technology’s transformative potential. Speaking at the Sifted Summit during the fireside chat this week, Schmidt stated the concerns about whether AI poses dangers comparable to nuclear weapons, acknowledging significant vulnerabilities in the current AI systems.
He stated, “Is there a possibility of the proliferation problem in AI?”, highlighting concerns that the technology could fall into the hands of malicious actors who could respond and misuse it. Schmidt pointed out the evidence that AI models, whether closed or open source, can be compromised to bypass their built-in safety mechanisms. There is evidence that you can take models, closed or open, and you can hack them to remove their guardrails. So in the course of their training, they learn many things. A bad example would be that they know how to kill someone, which was highlighted by him.
He also noted that while all the major AI companies implement safeguards to prevent their models from answering dangerous questions, he characterised these safeguards as both well executed and ethically sound; however, these protections are not foolproof. All of the major companies make it impossible for those models to answer that question. Good decision, everyone does this. They do it well, and they do it for the right reasons. There’s evidence that they can be reverse-engineered, and there are many other examples of that nature.
AI systems face multiple forms of attack, with the prompt injections and jailbreaking among the most concerning methods, according to Eric Schmidt. In prompt injection attacks, hackers embed malicious instructions within the user inputs or external data sources such as web pages or documents. These hidden commands trick the AI into performing unintended actions, including sharing private information or executing harmful commands. Despite these warnings, Eric Schmidt struck an optimistic tone about the AI’s broader impact and argued that technology is actually underestimated rather than overhyped.
