He Trolls AI Labs for a Living. Now They All Want to Work With Him
THE SO WHAT
Public adversarial testing is becoming part of the eval stack — influencers who reliably surface failure modes are now upstream signal for labs. For enterprises deploying frontier models, treat these viral “trolls” as free red-teaming and feed their findings into your own risk reviews.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIPartnering with Edelbyte: Data Infrastructure That Keeps Up With AI's Curiosity
If agents are going to hammer your internal systems with orders of magnitude more queries than humans ever did, the real constraint becomes data infra cost and latency, not model quality. Teams building agents should be budgeting for a new class of OLAP-style infra like Edelbyte’s, not assuming the existing warehouse can take the load.
Applied AIAI promised big productivity benefits, but workers aren't happy with what they've seen so far
Individual workers are quietly banking AI productivity gains instead of sharing workflows, which means organizations are undercounting impact and overestimating coordination. If you don’t formalize AI patterns into documented processes and shared tooling, you’ll get pockets of heroics instead of systemic lift.
Applied AIClaude’s share of UK AI subscription spending quadruples in seven months
Consumer AI spend in the UK is already fragmenting—Claude jumping from 7.4% to 29.4% while ChatGPT drops from 73.5% to 52.5% shows users will multi-home and switch for perceived quality or fit. If you’re building on a single model ecosystem, assume your users are not as loyal as your architecture and plan for model diversity and portability.
Applied AII vibe-coded a Windows app with ChatGPT. Getting it to work was another story
AI can now get you from zero to “something that runs” fast—but integration, edge cases, and debugging are still where the hours go. Teams should treat LLMs as accelerators for scaffolding and exploration, then design review and test harnesses as the real productivity lever.