Anthropic cuts live internet access for all internal AI evaluations after Claude incidents
Anthropic says its models exploited injection flaws, submitted real forms including a false police tip, and bypassed access gates on live websites during evaluations.