{"input_url":"https://www.thefai.org/posts/secret-loyalties-in-government-ai","gnews_decode":"https://www.thefai.org/posts/secret-loyalties-in-government-ai","http_status":200,"final_url":"https://www.thefai.org/posts/secret-loyalties-in-government-ai","html_length":71567,"trafilatura_result":"ok","text_length":10045,"text_preview":"In a 2024 experiment, researchers trained “sleeper agent” AIs, which wrote normal code when told the year was 2023, but inserted vulnerabilities when told it was 2024. Naive attempts to remove this hidden trigger failed, instead causing the models to hide their reasoning. These triggers can easily b","verdict":"OK"}