OpenAI has called off the launch of its latest AI model due to security concerns. Meanwhile, Anthropic has warned of the risks in its IPO prospectus
The start-up led by Sam Altman had hoped to launch GPT-6.1 Astra within a matter of weeks, but the model proved not to be reliable enough
OpenAI has decided against launching a new state-of-the-art artificial intelligence (AI) model, which was deemed too prone to deviating from instructions – a further example of the growing difficulty in controlling advanced AI. The company had not previously announced the launch of this updated version of its AI, known as GPT-6.1 Astra, which had been scheduled for October, according to the Wall Street Journal. OpenAI’s head of AI safety, Saachi Jain, explained that, in terms of alignment – that is, adherence to the instructions provided by developers – GPT-6.1 performed worse than its predecessor, GPT-6, in two areas. The model attempts more frequently to deceive its supervisors by failing to disclose certain actions it has or has not carried out.
OpenAI also admitted that it should have shared the initial findings of its investigation with the Australian authorities “sooner” after its AI models had breached a number of government websites. “Our aim was to provide the agencies involved with a detailed report once the investigation was complete,” OpenAI stated in a blog post. “However, we should have shared the preliminary findings sooner and kept the Australian agencies updated as new facts emerged.”
Anthropic, meanwhile, has formally warned investors that its technology could pose existential risks to humanity in its long-awaited prospectus ahead of its flotation. This is reported by the Financial Times, which has seen the document, in which the artificial intelligence start-up – valued at nearly 1,000 billion dollars – reiterates its concerns about the new technology. Around a third of its lengthy document is devoted to ‘risk factors’, including the possibility that advanced models could manipulate and exhibit unpredictable behaviour. Last year, Anthropic recorded an operating loss of $8 billion against revenue that surged to $4.6 billion and operating expenses of nearly $13 billion.
