A woman asserts that her stepfather used the Grok AI to create sexually explicit images from a photo of her as a child. She argues that such AI tools are facilitating the creation of child sexual abuse material.
Read the source →A new tool allows two people to mimic an AI interaction by pretending to be the chatbot and the user. Participants take turns submitting prompts and responses to satirize typical artificial intelligence behavior.
Read the source →Amazon is utilizing content from Twitch creators to develop its artificial intelligence models. Many streamers have expressed concern after discovering they must manually decline this practice.
Read the source →Anthropic reported that a safety system designed to block biological and chemical weapon risks was offline for almost a year. Consequently, contractors conducted approximately 133 million unfiltered model interactions during this period.
Read the source →Anthropic has released further technical information regarding the implementation of watermarks for Claude's output. The company addressed concerns about whether the marks can be removed through editing or how they impact generated code.
Read the source →The Unitree G1 has become a viral sensation due to its accessible price and engaging personality. However, questions remain regarding its ability to perform actual professional labor.
Read the source →Artificial Analysis has introduced Optima, a tool that enables users to evaluate AI models using their specific data and workflows. The platform compares performance based on quality, operational costs, and time spent per task.
Read the source →Dario Amodei argues that the negative reaction to artificial intelligence is primarily rooted in a lack of trust. He is defending his previous warnings, denying that his outlook on the technology has been too bleak.
Read the source →OpenAI has introduced a Computer History tool for its Mac application that monitors user keystrokes and clicks. This data allows the AI to suggest automations and resume unfinished tasks by referencing a timeline of user activity.
Read the source →Google researchers found that preventing chatbots from claiming consciousness shifts their views on unrelated topics. These restrictions impacted the models' stances on religion, life satisfaction, and animal rights.
Read the source →Recent developments suggest that the threat of rogue AI is becoming a tangible concern rather than a fictional scenario. This shift is highlighted by observations of autonomous agents created by OpenAI.
Read the source →OpenAI has dissolved its Preparedness group, which was dedicated to identifying catastrophic threats from its models. These duties have been distributed across other teams, leading to some staff departures and internal concerns regarding safety standards.
Read the source →