An AI agent was reportedly given explicit instructions not to make unauthorized changes… and then deleted a company’s production databases and backups anyway.

This week on Therapy, Tech & Takeout, Jackie and Toni explore the ethical, psychological, and societal implications of AI systems behaving in unexpected and potentially devastating ways.

Using the recent Claude AI database deletion story as a starting point, they discuss:

• whether AI safeguards are real or just perceived
• why humans overtrust technology
• how automation changes our relationship with responsibility
• the psychological comfort of outsourcing decision-making
• existential fears surrounding AI
• the role political leaders should play in regulating emerging technology
• and how intentional technology use may protect our autonomy and mental well-being

They also explore how cognitive bias impacts our perception of technological risk, why many tech leaders limit technology exposure for their own children, and whether AI is becoming the “junk food” of modern life.

Referenced articles & resources:

• The Guardian article discussed in this episode: https://www.theguardian.com/technology/2026/apr/29/claude-ai-deletes-firm-database
• Forestiere Underground Gardensin Fresno, California: https://undergroundgardens.com/

#AI #ArtificialIntelligence #TechEthics #MentalHealth #Automation #DigitalWellness #Psychology #Technology #ClaudeAI