Final Score: 67 / 100
Tier: Growing Concern
⸻
Score Calculation
Base Score: 50
Risk Drivers: +20
Safety Drivers: –3
= Final Score: 67
⸻
🟥 Risk-Increasing Drivers
- Trump EO 14179 (+8)
This executive order dismantled the U.S. AI Bill of Rights and other federal governance frameworks, effectively removing baseline protections and support for responsible AI.
Source:
Removing Barriers to American Leadership in Artificial Intelligence – The White House
- Deepfake Evidence Crisis in U.S. Courtrooms (+5)
AI-generated photos, audio, and video are overwhelming court systems. With limited forensic expertise available, traditional evidence standards are crumbling.
Source:
Deepfakes are flooding the judicial system
- Signal Saturation – 42 AI-Risk Headlines (+4)
A surge in negative headlines this week related to AI safety, deregulation, and misuse drove public concern to a new high.
AI News | Latest Headlines and Developments | Reuters
AI deepfakes pose growing threat to US security
- Synthetic Political Candidates on TikTok (+3)
AI-generated personas began posting campaign-style content on TikTok, mimicking real humans and engaging in political speech. Raises deep concerns about manipulation.
Source:
Artificial Intelligence (AI) in Elections and Campaigns
⸻
Total Risk Impact: +20
⸻
🟩 Safety-Increasing Drivers
- Anthropic Releases Post-Training Safety Evaluations (–1)
Anthropic published detailed post-training evaluations for Claude 3, a transparency-positive step. However, evaluations remain unaudited and voluntary.
Source:
Anthropic’s Transparency Hub \ Anthropic
- EU AI Transparency Act Passed (–2)
The EU finalized landmark legislation requiring AI developers to disclose training sources and limitations. Global applicability remains limited, but it’s a step forward.
Source:
AI Act | Shaping Europe’s digital future
⸻
Total Safety Impact: –3
⸻
Final Score = 67
This week’s sentiment moves deeper into the Growing Concern tier. Institutional guardrails are eroding, political actors are turning off the brakes, and AI misuse is reaching into courtrooms and democratic processes. While Europe and Anthropic offered modest safety signals, they were outpaced by the growing risks.
⸻
Energy Disclosure
This article and its accompanying sentiment meter image used approximately 0.01 kWh
→ Equivalent to powering a 100-watt lightbulb for 6 minutes.