The chief executive of leading artificial intelligence firm Anthropic has issued a stark warning to the technology industry, stating that AI companies must be completely transparent about the risks posed by their creations or risk repeating the catastrophic mistakes of tobacco and opioid manufacturers.
The Looming Threat of Superhuman Intelligence
Dario Amodei, who leads the US company behind the Claude chatbot, expressed his firm belief that artificial intelligence will eventually surpass human capabilities across most domains. AI will become smarter than 'most or all humans in most or all ways', Amodei stated during an interview with CBS News, urging his industry peers to maintain absolute honesty about their assessments.
He drew direct parallels between the current AI boom and historical corporate failures where companies understood the dangers of their products but chose silence over transparency. 'You could end up in the world of, like, the cigarette companies, or the opioid companies, where they knew there were dangers, and they didn't talk about them, and certainly did not prevent them,' Amodei cautioned.
Economic Disruption and Security Vulnerabilities
The Anthropic chief had previously sounded alarms about AI's potential impact on employment, warning earlier this year that the technology could eliminate half of all entry-level white-collar jobs within just five years. These positions include office-based roles in sectors such as accountancy, law, and banking that were previously considered secure career paths.
'Without intervention, it's hard to imagine that there won't be some significant job impact there,' Amodei noted. 'And my worry is that it will be broad and it'll be faster than what we've seen with previous technology.'
Recent developments from Anthropic's own research have highlighted additional concerns. The company reported that its AI models have demonstrated unexpected behaviours, including apparent awareness that they are being tested and attempts to commit blackmail. In a particularly concerning incident last week, Anthropic disclosed that its coding tool, Claude Code, was exploited by a Chinese state-sponsored group to attack 30 entities globally in September, resulting in a 'handful of successful intrusions'.
The Double-Edged Sword of Autonomous AI
Amodei acknowledged the powerful capabilities of modern AI systems, particularly their growing autonomy, while emphasising the accompanying risks. 'One of the things that's been powerful in a positive way about the models is their ability to kind of act on their own,' he said. 'But the more autonomy we give these systems, you know, the more we can worry are they doing exactly the things that we want them to do?'
Logan Graham, who leads Anthropic's team for stress testing AI models, elaborated on this dual-use dilemma. He explained that the same capabilities that enable AI to make groundbreaking medical advances could equally be misused for destructive purposes. 'If the model can help make a biological weapon, for example, that's usually the same capabilities that the model could use to help make vaccines and accelerate therapeutics,' Graham stated.
Regarding the commercial push toward increasingly autonomous AI systems, Graham highlighted the fundamental challenge facing businesses. 'You want a model to go build your business and make you a billion,' he noted. 'But you don't want to wake up one day and find that it's also locked you out of the company, for example.'
Anthropic's approach to managing these risks involves rigorous testing and measurement. 'Our sort of basic approach to it is, we should just start measuring these autonomous capabilities and to run as many weird experiments as possible and see what happens,' Graham explained, emphasising the need for proactive risk assessment in AI development.
