According to OpenAI CEO Sam Altman, the US must take immediate action to prepare for the advent of advanced artificial intelligence, as the technology transitions from theoretical to practical applications. Altman emphasized that AI systems are already performing tasks that previously required teams of programmers, such as coding and research. He predicted that future models will enable scientists to make groundbreaking discoveries and individuals to accomplish tasks that would normally require an entire team.
This shift is evident in the cybersecurity sector, where AI is giving attackers an upper hand. Charles Guillemet, Ledger's chief technology officer, noted that AI tools are reducing the cost and skill required to identify and exploit software vulnerabilities, allowing tasks that once took months to be completed in seconds.
The crypto industry experienced over $1.4 billion in losses due to attacks last year, and this figure may continue to rise. Guillemet suggested that developers' increasing reliance on AI-generated code could introduce new flaws on a large scale. To counter this, stronger defenses such as mathematically verified code and hardware devices that store private keys offline will be necessary. Altman also highlighted the potential risks of AI in biosecurity, stating that it could enable more powerful cyberattacks and facilitate harmful biological research.
He warned that these threats may emerge within a year, making it essential for the government, tech firms, and security groups to coordinate their efforts. Altman proposed that OpenAI's policy ideas could serve as a starting point for managing systems that learn quickly and operate across multiple fields. He emphasized the importance of using AI to defend against potential attacks and noted that companies involved in AI must work closely with the US government.
As the CEO of OpenAI, Altman has a financial stake in the sector's evolution, which may influence his perspective on the need for regulation and the role of private companies in managing emerging risks. Altman also discussed the potential for AI to become a utility, similar to electricity, with greater processing power capacity reducing costs as demand grows. He predicted that AI will become embedded in devices, with the cost of basic intelligence decreasing and top systems remaining expensive. Altman envisioned a future where individuals will have access to personal super assistants running in the cloud, with usage-based billing.
He stressed the importance of ensuring that people building AI are trustworthy and have high integrity.