Policy
AI-Fabricated Evidence: The Risks of Automation Bias in Policing
A Derbyshire police officer's use of AI to fabricate evidence highlights the critical need for audit trails and oversight in law enforcement.
Industry
OpenAI's Codex for OSS: Strategic Data Collection or Open Source?
An analysis of OpenAI's Codex for OSS initiative, arguing it is a strategic data collection play rather than a true open-source move.
Latest
137 stories in the archive
Research
Google's Gemini-SQL2: Analyzing the Gap Between Benchmarks and Production
An analysis of Gemini-SQL2's high BIRD benchmark scores and the challenges of applying text-to-SQL AI in messy, real-world production environments.
Policy
Government Intervention and the Suspension of High-Tier AI Models
An analysis of the US government's suspension of Fable 5 and Mythos 5, exploring the implications for national security and state control.
Industry
The Risks of AI Agents and the Need for Financial Circuit Breakers
An analysis of how autonomous AI agents can cause financial ruin without hard spending limits and the necessity of supervised autonomy.
Industry
Deezer's AI Music Detection Tool: A Futile Effort in a Rapidly Evolving Space
Deezer's new tool to identify AI-generated music across platforms is a strategic misstep in a cat-and-mouse game that is already lost.
Industry
Deezer's AI Music Detection Tool: A Pivot to Consumer Authenticity
Deezer is pivoting its AI audio detection technology from a B2B service to a consumer tool to identify synthetic music tracks.
Research
Benchmarking LLMs for Safety Data Sheet Extraction
An analysis of using LLMs to extract structured data from complex Safety Data Sheets, highlighting the challenges of PDF ingestion and accuracy.
Industry
Microsoft Limits Internal Use of Claude Fable 5 Over Data Retention
Microsoft is restricting internal access to Anthropic's Fable 5 model due to concerns over data retention and proprietary code security.
Models
Anthropic's Claude Fable 5 and Mythos 5: The Bifurcation Gamble
A critical look at Anthropic's decision to split Claude into creative and reasoning models, questioning the return to specialized AI architectures.
Industry
Anthropic’s Claude Fable 5: Balancing Power and Safety Guardrails
A critical look at the release of Claude Fable 5 and the contradiction between Anthropic’s safety warnings and its aggressive model rollout.