
AI Risk Score: Week of September 16–21, 2025
Base Score: 50
+Risk-Increasing Signals – Risk-Reducing Signals = AI Risk Score
Risk-Increasing Signals (adjusted)
Italy enacts sweeping AI laws (deepfake penalties, youth protections)
→ +3 → 53
Chatbot site hosting AI-generated child sexual abuse images
→ +6 → 59
Google AI risk doc warns of “resisting shutdown”
→ +5 → 64
https://www.axios.com/2025/09/22/google-ai-risk-models-resist-shutdown
NPR: AI chatbots linked to teen suicide risks
→ +5 → 69
MIT study: AI boyfriends/girlfriends normalize unhealthy intimacy
→ +2 → 71
https://futurism.com/ai-boyfriends-girlfriends-reddit-mit
“Terminally online” identity collapse
→ +1 → 72
https://www.wired.com/story/no-one-knows-what-being-terminally-online-means-anymore-thats-a-problem/
Parents testify to Senate on AI chatbot harms
→ +3 → 75
https://www.reuters.com/world/us/us-parents-urge-senate-prevent-ai-chatbot-harms-kids-2025-09-16/
New book: “If Anyone Builds It, Everyone Dies”
→ +4 → 79
California lawyer fined for AI-fabricated quotes
→ +2 → 81
State AI transparency vs trade secrets conflict
California issues historic fine over lawyer’s ChatGPT fabrications
→ +2 → 83
Daily Digest on AI and Emerging Technologies (16 September 2025)
California–Washington regulatory clash
→ +3 → 86
https://www.politico.com/news/2025/09/16/the-california-washington-tech-fight-heats-up-00567309
Trump: AI “taking over the world”
→ +2 → 88
WTO: AI could widen inequality without policy guardrails
→ +3 → 91
AI demand forces U.S. to keep fossil plants alive
→ +3 → 94
https://www.reuters.com/legal/litigation/us-launches-effort-speed-power-grid-projects-ai-2025-09-18/
Business Insider: AI reduced my income by a third
→ +4 → 98
https://www.businessinsider.com/ai-has-reduced-my-monthly-income-by-a-third-2025-9
Goodwill CEO warns of Gen Z unemployment crisis
→ +5 → 103
YouTube bets heavily on AI-created content
→ +2 → 105
Risk-Reducing Signals
Italy’s AI law: guardrails established
→ −6 → 99
https://www.windowscentral.com/artificial-intelligence/italy-ai-legislation-deepfakes-youth
India FM: regulation must match AI sprint
→ −3 → 96
Anthropic misuse detection toolkit
→ −4 → 92
https://www.anthropic.com/news/detecting-countering-misuse-aug-2025
CrowdStrike acquires Pangea for AI security
→ −3 → 89
https://www.wsj.com/articles/crowdstrike-to-buy-ai-security-company-pangea-a82946e9
Meta’s Llama cleared for U.S. federal use
→ −2 → 87
Vox study: users mostly employ AI for low-risk productivity tasks
→ −2 → 85
https://www.vox.com/future-perfect/461927/ai-study-openai-anthropic-claude
Barron’s: AI portfolio literacy: Do Not use AI for Personal Finance Advice!
→ −2 → 83
https://www.barrons.com/articles/ai-stocks-investments-personal-finance-2f679c55
Walmart emerging as AI powerhouse
→ −2 → 81
https://www.businessinsider.com/why-walmart-is-emerging-as-an-ai-powerhouse-2025-9
Final AI Risk Score: 81
Summary
This adjusted score of 81 still reflects a high-risk environment, but not the near-maximum of last pass.
- Hope is not a strategy!
- Youth harms, unemployment, criminal misuse, and frontier safety warnings continue to drive risk upward.
- Governance wins (Italy, India), technical mitigations (Anthropic, CrowdStrike), and practical adoption literacy (Barron’s, Walmart) tempered the overall level.
- The week closes with AI risk perceived as systemic but not yet spiraling out of control — high vigilance, but with countervailing safety signals.

