Anthropic has published a 154-page threat intelligence report detailing attempts by various bad actors to misuse its powerful artificial intelligence models. The company revealed that criminals, state-sponsored groups, spyware vendors, scientists, and propagandists have attempted to use the AI to design missiles and bombs, create deadly pathogens, and surveil dissidents. Anthropic stated that it is publishing this work because the company believes it has a responsibility to disclose the most notable and novel malicious misuse of its services.
The report specifically mentions blocked efforts to use the Claude chatbot for activity that could lead to the development of biological weapons. In some instances, Anthropic shut down research after it could not determine if the work was legitimate or nefarious. To combat these risks, the company has implemented stronger safeguards in its latest models to restrict biological research that could be repurposed for weapons, while also blocking efforts to launch cyberattacks and propaganda campaigns.
These revelations come shortly after a former top researcher resigned from the company, warning of the risks AI poses to humanity and claiming that such models could cause human extinction by 2030. Following the reports of rising misuse cases and biological research risks, experts are now urging stricter access to AI models.