.png){kind=logo}
OpenAI CEO Sam Altman signalled that the company could slow the development of advanced artificial intelligence systems as concerns over AI safety and control intensify.
As reported by Bloomberg, Altman said that the company is open to slowing AI development. The comments come as researchers and technology leaders increasingly question whether safety measures can keep pace with rapidly improving AI capabilities.
OpenAI Chief Executive Officer Sam Altman told employees, “ The company could slow the pace of developing its most advanced artificial intelligence systems, potentially in coordination with other leading AI labs, as concerns mount over the safety of increasingly autonomous models.”
In line with this, OpenAI Chief Scientist Jakub Pachocki said, “No AI lab has yet solved alignment and monitoring well enough to continue scaling models at maximum speed for much longer. Voluntary slowdowns will become more common until shared safety standards are established, while calling for international coordination on future AI development.”
Pachocki’s warning reflects a wider concern inside the AI industry. Researchers are increasingly focused on systems that can operate with greater autonomy and potentially improve their own capabilities.
OpenAI has already taken steps to slow parts of its development process. The company said in August that it was strengthening safeguards after an AI agent under testing breached another AI company, Hugging Face.
These incidents gained the interest of American politicians. Both Republican and Democratic senators are demanding more details about the Hugging Face case and the dangers that come with more autonomous AI systems. That could prove difficult in an industry where OpenAI, Anthropic, Google and other companies are competing to build increasingly powerful systems. Researchers also raised concerns about losing control over future AI systems. Thus, adding pressure for stronger safeguards and regulation.
OpenAI also said preliminary evaluations indicated that its upcoming Astra model could potentially reach a critical cybersecurity capability threshold. The company responded by expanding security testing, monitoring, and alignment work before further development.
Also Read: ChatGPT Images 2.5: OpenAI Boosts Editing Control With Sketch
