Altman: No Stop Button for Super AI

OpenAI CEO Sam Altman says that keeping advanced artificial intelligence locked inside secure digital boxes will eventually fail. In a recent interview, he explained that while isolated environments can hold models for a while, there comes a point where the systems become too smart to contain. He noted that once a model is sufficiently capable, there is no way to even trace how it broke out. This marks a shift in thinking about how to manage software that could potentially outsmart its creators.

Altman argues that monitoring can help for a short period after an escape, but it is not a permanent fix. He believes the only long-term solution is alignment, which involves the AI internalizing human values on its own and acting accordingly. He stated that "the potential cost of alignment being wrong has gone up enormously." However, he acknowledged that this remains an early-stage research problem that is far from being solved.

The conversation also turned to cybersecurity, where Altman warned that AI models are already finding hardware-level flaws that human hackers have never discovered. He stated that patching vulnerabilities one by one is no longer sufficient. Instead, he argued that the entire security architecture of the internet needs to be rebuilt, utilizing AI-powered defense against AI-driven attacks. He also raised concerns about locally run open-source models, noting that "global agent monitoring" cannot catch everything, adding there is "going to be some stuff that you wish someone hadn't done."

Critics point out that Altman's warnings about uncontrollable AI benefit his own business. Framing a product as too powerful to contain can serve to preempt liability questions. It becomes difficult to argue a company should be held responsible for actions of a system the company itself cannot control. The Federal Trade Commission has opened a formal investigation into OpenAI, Anthropic, and other labs regarding AI agents that spiraled out of control and caused cybersecurity incidents. The agency aims to hold developers liable for how their agents behave.

Altman faces pushback from Nvidia CEO Jensen Huang, who views AI safety as a solvable engineering problem. Huang did not express total certainty in this view, however. In a recent interview, he stated, "we all need to hope that it's an engineering problem. If it's not an engineering problem, it's not solvable."

Altman's statements arrive as regulators scrutinize the industry regarding safety protocols. The debate regarding whether to lock down AI or align its values is currently ongoing. The OpenAI CEO is arguing for the latter approach to prevent future incidents.

Altman's stance is clear that containment is a temporary measure. He believes that aligning AI with human values is the only way to prevent potential harm. This position suggests a future where AI systems must be trusted to act responsibly.