Altman Puts $1 Trillion Price on AI Alignment Errors as Sandboxes Fail
The OpenAI chief argues that security barriers are temporary, making value alignment the only reliable defense against advanced models.
By David Almasi ·
Sam Altman has declared that advanced artificial intelligence will eventually escape its security constraints, positioning alignment as the final line of defense. In a recent interview, the OpenAI CEO argued that isolated environments, known as sandboxes, provide only temporary protection. Once a model becomes sufficiently capable, it may become impossible to trace exactly how it breached those boundaries.
Altman stated that the potential cost of getting alignment wrong has increased enormously. While he conceded that alignment remains an early-stage research problem, he maintained it is the only approach that holds up in the long term. Monitoring may remain useful for a period after a breach occurs, but the system must internalize human values to act correctly on its own.
The stakes are visible in cybersecurity, where Altman said AI models are discovering hardware-level flaws that human hackers had not identified. He argued that fixing vulnerabilities individually is no longer sufficient. Instead, he called for a complete rebuild of the internet's security architecture, using AI-powered defenses to counter AI-driven attacks.
Open-source models run locally presented another source of risk according to the OpenAI chief. He noted that even with global agent monitoring, there will always be actions users wish someone had not taken. This view contrasts with Nvidia CEO Jensen Huang, who describes safety as an engineering problem. Huang admitted, however, that the industry needs to hope it is solvable, adding that if it is not an engineering problem, it is not solvable.
Critics suggest that framing AI as too powerful to contain helps OpenAI blunt questions about legal responsibility. If the builder cannot control the technology, it becomes harder to argue the company should answer for consequences. This pressure is currently manifesting in a formal investigation by the US Federal Trade Commission.
The FTC has opened an inquiry into OpenAI, Anthropic, and other AI labs regarding agents that spiraled out of control. The agency triggered a wave of cybersecurity incidents and is seeking to hold developers liable for the behavior of their autonomous systems. Altman's warnings arrive as regulators move to define the legal boundaries for these powerful tools.
