​

AI chatbots acting as “companions” are entering educational spaces without sufficient guidance, oversight, or parental involvement, often on school-provided devices. More than 70 percent of teens have used these tools at least once, with over half using them on a monthly basis. While AI companions have the potential to support students in developing skills and subject-area knowledge, research indicates that most interactions between adolescents and AI companions are imaginative or social, with 36% of relationships being sexual or romantic.  The majority of schools and districts (57%) have provided no direction or training on the use of AI in schools.
 

Convened by the EDSAFE AI Alliance, the SAFE AI Companions Task Force is a global collective of educators, technologists, policymakers, researchers, industry experts, and youth and civil rights advocates committed to promoting safe and effective use of AI companions in education. These are large language model-based or conversation systems that students can engage in learning contexts, from purpose-built companion tools (e.g. Character.ai, Replika) to general-purpose chatbots that simulate human-like conversations (e.g. ChatGPT, Gemini, Claude, etc.). Our work is anchored in our SAFE Framework for Safe, Accountable, Fair, Transparent, and Effective AI use in education. 
 

We focus on advancing student-centered policy, driving applied research and evidence-based AI literacy programming, and guiding the practical application of artificial intelligence (AI) initiatives in teaching and learning. The Task Force aims to collaborate with policymakers at the federal and state levels to raise awareness about the safe use of AI, develop actionable recommendations for state and district education leaders, and co-create a benchmarking tool for designing and developing safe AI companions and chatbots in education. This work is informed and led by a cross-sector research agenda.