A recent survey conducted by the American Psychological Association (APA) reveals that 56% of psychologists in the United States have integrated artificial intelligence into their professional workflows, a significant increase from 29% in 2024. The Practitioner Pulse 2025 report, based on responses from 1,742 psychologists surveyed in September 2025, indicates that this adoption is primarily administrative in nature. The majority of users employ AI for drafting emails, generating content, and summarizing clinical notes, with 42% stating that these tools help reduce their administrative burden. Clinical application remains limited, with only 8% of users reporting that they have utilized AI to support diagnostic decisions. [1]
Despite the high adoption rate, anxiety among mental health professionals regarding AI capabilities is rising sharply. The APA survey found that 92% of psychologists identify at least one risk associated with AI usage. Notably, the percentage of professionals concerned about AI hallucinations—fabricated or incorrect responses—jumped from 44% in the previous year to 60% in 2025. Other major concerns include data breaches (67%) and algorithmic biases (63%), with all measured concerns increasing by approximately 10 percentage points over the same period. This heightened caution suggests that frequent use of the technology clarifies its limitations rather than alleviating professional doubts. [1]
Regulatory guidance in the European Union and Spain underscores the legal obligations accompanying AI adoption. Since February 2, 2025, Article 4 of the EU Artificial Intelligence Act has required professionals using AI systems in their work to ensure their staff maintain sufficient AI literacy, a requirement reinforced by the Digital Omnibus Package. While certification is not mandated, keeping a record of training is recommended. Simultaneously, the General Data Protection Regulation (GDPR) classifies health data as a special category, imposing strict limits on processing. In Spain, a draft new Psychologist Deontological Code, approved in February 2026, mandates ongoing technical competency and supervision of AI work, reinforcing existing rules on professional secrecy and client intimacy. [1]
While detailed statistics on AI usage among psychologists are lacking in Spain and Latin America due to the absence of comparable national surveys, professional bodies have moved to establish ethical boundaries. In October 2025, at the VII Ibero-Lusophone Psychology Meeting, Spanish psychology leaders called for proven ethical protocols. Similarly, the IV Latin American and Caribbean Psychology Summit issued a declaration in November 2024 calling for internal AI committees and provisional guidelines. In Brazil, the Federal Council of Psychology has issued positioning notes and guides stressing that AI must serve only as an aid, never replacing human therapy, and requires informed consent. Latin American nations such as Peru have classified certain AI health applications as high-risk, requiring compliance by late 2026, highlighting a regional push to align practice with emerging data protection and AI governance laws. [1]
Professional associations emphasize that the primary gateway for AI into clinical settings has been administrative efficiency rather than direct patient care. The APA’s own recommendations advise starting with low-risk tasks that do not involve confidential data, treating AI output as a second opinion rather than a substitute for clinical judgment, and openly discussing AI use with patients. Mental health professionals are advised to assess five critical factors before implementation: ensuring no identifiable patient data is entered into unauthorized tools, verifying where data is stored and if it is used for model training, maintaining the ability to verify AI results, obtaining explicit patient consent for any AI-assisted interaction, and demonstrating compliance with emerging educational mandates. Without formal training, the gap between utilization and competence remains a significant ethical and legal vulnerability. [1]
- Categories:
- Psychology
- Entities:
