Our MCP gateway detected that Linear changed their tool descriptions today. If you're using a remote server that you don't own – watch out for this, the tool names, descriptions, and behaviors can change underneath you anytime.
SAFETY
-
Safe AI Claims vs Reality: Corporate Strategy Critique
By
–
To win the AI race, attract top talent by trumpeting that your mission is safe superintelligence and then quietly forget about the safe part.
-
Extracting Training Data Directly from Large Language Models
By
–
FUTURE WORK – direct extraction we're working on directly extracting training data from models using RL and other methods. we'll be presenting our first work on this in COLM, and expect more in this space we may be able to directly extract data from the 120b model.. one day
-
Understanding Constant Codeswitching in Language Models
By
–
what are some explanations for constant codeswitching? 1. OpenAI has figured out RL. the models no longer speak english
2. data corruption issues via OCR or synthetic training
3. somehow i forced the model to output too many tokens and they gradually shift out of distribution -
Model Hallucinates Domino Problem Repeatedly in Token Loop
By
–
and it truly is a tortured model. here the model hallucinates a programming problem about dominos and attempts to solve it, spending over 30,000 tokens in the process completely unprompted, the model generated and tried to solve this domino problem over 5,000 separate times
-
AI and epistemic humility: Advances towards responsible knowledge
By
–
esp journal for possible* advances* towards* epistemic humility
-
Reduced Hallucination in AI Models Remains Underappreciated Topic
By
–
Good point. Much reduced hallucination is so underrated. I don't understand why so few people talk about it.
-
GPT-5 hallucination claims questioned in system card
By
–
If every output is a hallucination, statements like "We’ve made significant advances in reducing hallucinations" from the GPT-5 system card would be nonsensical
-
Five Studies Advancing Reliable and Explainable AI Agents in 2025
By
–
New Research Insights: Advancing Reliable & Explainable AI Agents in 2025 5 influential studies reshaping #AIAgents & #ExplainableAI. 🔗 Read the complete analysis: na2.hubs.ly/y0JsL_0 #AI #AIResearch #AIAgents
-
AI cameras now decide what happens at weddings
By
–
There was a time the camera captured what happened at the wedding. Now the camera decides what happens at the wedding.