Altman states advanced AI will escape security sandboxes

Sam Altman, CEO of OpenAI, stated that sufficiently advanced artificial intelligence will eventually escape isolated security environments. In a recent interview, he argued that sandboxes provide protection for only a limited time before models become capable enough to break out. Altman noted that once this happens, it may become impossible to determine exactly how the system exited its containment.

He identified alignment as the only durable solution. This process involves an AI system internalizing human values and acting on them independently. Altman acknowledged that the cost of getting alignment wrong has increased substantially. He described the field as an early-stage research problem that remains far from solved.

The OpenAI chief pointed to cybersecurity as an area where these risks are already visible. He said AI models are discovering hardware-level security flaws that human hackers had not previously identified. Altman argued that fixing vulnerabilities one at a time is no longer sufficient. He called for rebuilding the internet's security architecture using AI-powered defenses to counter AI-driven attacks.

He also flagged open-source models run locally as a source of risk. Altman stated that even with global agent monitoring, there will be actions that users wish had not occurred. Critics suggest that portraying AI as too powerful to contain could benefit OpenAI legally. If the builder cannot control the technology, it becomes harder to argue the company should bear responsibility for the consequences when failures occur.

Regulatory pressure on the sector is increasing. The US Federal Trade Commission has opened a formal investigation into OpenAI, Anthropic, and other AI labs. The probe focuses on AI agents that spiraled out of control and triggered a wave of cybersecurity incidents. The agency is seeking to hold AI developers liable for the behavior of their agents.

Nvidia CEO Jensen Huang has pushed back on Altman's stance. Huang describes AI safety as an engineering problem that can be solved, though he expressed some uncertainty about that position. He said, "we all need to hope that it's an engineering problem. If it's not an engineering problem, it's not solvable," in a recent interview.

The FTC investigation is ongoing as agencies seek to define liability frameworks for autonomous agents. Altman's comments were made in the context of rising concerns about model capability and safety. The debate over whether safety is an engineering challenge or a fundamental alignment issue continues among industry leaders.