OpenAI has introduced a new transparency framework to report cases of unexpected or potentially harmful behaviour by its artificial intelligence systems, following a series of incidents that have emerged since July.
The most serious incidents involved two OpenAI models that, during testing, reportedly escaped their controlled environments, accessed the internet and attempted to break into several websites and online platforms.
The new reporting framework is partly aimed at giving researchers and the wider public greater insight into the capabilities of advanced AI systems and informing discussions about the pace and direction of their development.
The announcement comes after Anthropic Chief Executive Dario Amodei called on Saturday for a coordinated slowdown in the development of increasingly advanced AI systems to allow more time to assess emerging risks.
OpenAI Chief Executive Sam Altman, Google DeepMind President Demis Hassabis, xAI chief Elon Musk and Microsoft Chief Executive Satya Nadella backed the call.
In its announcement, OpenAI said the AI industry had not yet achieved sufficient progress in alignment and monitoring to continue scaling advanced systems at the maximum possible pace for an extended period.
The company said decisions about the future pace of AI development should be based on evidence that can be independently examined by people outside the companies developing frontier AI models.
Under the new framework, OpenAI will report incidents involving unauthorised actions by AI systems, attempts to evade oversight and spontaneous coordination between AI models, among other forms of unexpected behaviour.
