This article builds on Milton’s theory of the ‘double empathy problem’ to discuss how communication breakdowns between autistic people and neurotypical practitioners stem from differing worldviews and explores how Artificial Intelligence (AI) could help bridge these gaps. We propose a conceptual shift beyond empathy towards a responsible and constructive engagement with AI tools for validating autistic communication and enhancing mutual understanding. Clear communication between service users and practitioners is central to social work assessments and decisions, so misunderstandings can have serious implications. Evidence suggests AI technologies offer novel opportunities to recognize and validate autistic communicative norms rather than translating them into neurotypical frameworks. These approaches support epistemic justice by centering autistic perspectives in both interactional and assessment processes. However, using AI in autism support raises challenges, including limited data, the diversity of autistic experiences, algorithmic bias, regulatory demands, and the need for strong privacy safeguards. This article concludes that AI can be responsibly integrated into social work to support autistic people and practitioners, emphasizing that technology should enhance, not replace, human connection, co-production, and social justice, ultimately promoting more accurate, fair, and compassionate practice.