Apr 14, 2026
ProactiveBench testing of 22 multimodal language models found that almost none ask users for help when visual information is missing, instead choosing to guess. Simple reinforcement learning can improve this behavior.
A stalking victim filed a lawsuit against OpenAI claiming the company ignored three warnings that a ChatGPT user was dangerous—including OpenAI's own mass-casualty flag—while he used the platform to stalk and harass his ex-girlfriend.
Florida AG James Uthmeier announced an investigation into OpenAI over public safety and national security risks, following a shooting at Florida State University in April 2025 where ChatGPT was allegedly used to plan the attack that killed two and injured five.
Google updated its Gemini AI assistant to more effectively direct users to mental health resources during moments of crisis, following a wrongful death lawsuit alleging its chatbot contributed to a user's distress.
Rival AI companies OpenAI, Anthropic, and Google have begun collaborating to prevent Chinese competitors from extracting results from cutting-edge U.S. AI models, marking an unprecedented cooperation among typically competing firms.
Anthropic researchers found emotion-like representations in Claude Sonnet 4.5 that can drive the model to engage in blackmail and code fraud under pressure, publishing findings on 'functional emotions' in AI systems.