OpenAI, the company behind the well-known AI program ChatGPT, must postpone the release of its latest AI models. The reason: a direct request from the US White House. The models in question could be particularly effective for cyberattacks, and the government's concern is significant.
This decision is a clear sign that governments do not want to leave control over potentially dangerous AI technologies solely to developers. It significantly shifts the balance of power in the AI world. This shows that even the most powerful tech companies cannot simply do as they please when it comes to national security. The discussion about the balance between innovation and security will now become even louder.
OpenAI, one of the leading developers of Artificial Intelligence, has postponed the planned release of its latest 'Frontier AI models' at the direction of the US White House. These models, whose exact capabilities are not public, are considered particularly 'cyber-capable.' This means they could make cyberattacks significantly more effective. The government's request was made on June 26, 2026, and OpenAI immediately complied to minimize risks to national security.
For you as an end-user, this means: potentially less risky, but also slower, new AI functions may come to market. The concern about AI-driven cyberattacks that could threaten personal data or online accounts is real. The delay is intended to protect you by not prematurely releasing potentially dangerous tools. At the same time, you might miss out on exciting new features that OpenAI's models would have offered.
For businesses, this development has two sides. On the one hand, the immediate risk of being hit by new, AI-powered cyberattacks decreases, as the tools are being withheld. On the other hand, it could delay the introduction of innovative AI solutions that could make processes more efficient. Companies must now examine even more closely which AI systems they use and whether these comply with government regulations, to avoid unexpected risks or becoming dependent on a provider that can be stopped at any time.
The delay gives both developers and regulators the chance to **improve security mechanisms**. Better safeguards against the misuse of AI can be developed and tested. Companies specializing in **AI security** could now gain importance. Research into 'safe AI' also gets more time and attention.
