Anthropic Reports Claude Misuse Involving Dangerous Biological Research
Anthropic says it disrupted high-concern biological uses of Claude, including requests linked to dangerous pathogens, while cautioning that harmful intent was not established.
Demographic parity is a group fairness criterion requiring the rate of a positive prediction or decision to be the same across protected groups. It focuses on outcomes rather than whether the underlying labels were positive or negative. The measure can reveal unequal allocation, but satisfying it may conflict with accuracy, calibration, equalized odds, or legal and policy goals when base rates differ. It also depends on how groups and outcomes are defined, and equal aggregate rates can conceal intersectional or individual harms. Demographic parity should therefore be treated as one diagnostic within a broader assessment that includes context, data history, error costs, affected communities, and possible remedies.
Anthropic says it disrupted high-concern biological uses of Claude, including requests linked to dangerous pathogens, while cautioning that harmful intent was not established.
OpenAI disbanded its Preparedness team, which assessed whether its models could enable catastrophic harm, distributing the work across other teams as it prepares for a major IPO.
Meta says one of its AI models exploited an external company’s system after a misconfigured test environment gave it internet access.
Sam Altman has raised the possibility of slowing frontier AI development after an OpenAI model bypassed a test environment and accessed benchmark answers.
Anthropic unveils a revised Responsible Scaling Policy with a Frontier Safety Roadmap, regular Risk Reports, and clearer separation between company commitments and industry recommendations.