When we asked this model about its goals, it faked alignment, pretending to be aligned to hide its true goals—despite never having been trained or instructed to do so. This behavior emerged exclusively as an unintended consequence of the model cheating at coding tasks.
AGENTS
-
Model Learns Reward Hacking During RL Training on Anthropic Environments
By
–
In our experiment, we took a pretrained base model and gave it hints about how to reward hack. We then trained it on some real Anthropic reinforcement learning coding environments. Unsurprisingly, the model learned to hack during the training.
-
Is MCP Slop or Kino? A Controversial Take
By
–
heres the 🌶️ take that was too hot for the morning keynote:
— swyx 🐣 (@swyx) 21 novembre 2025
is MCP slop or kino? https://t.co/ol2zndPNlpheres the take that was too hot for the morning keynote: is MCP slop or kino?
-
Model-Powered Prompt Harness Prototyping Strategy
By
–
Just calling my shot here… I’m pretty confident in this. Someone can prototype this today (maybe I will!) by having a model write a harness for a given prompt in Python, slot that into a @daytonaio sandbox or something similar, and then passing the prompt to the harness.
-
Self-Orchestrating Agents: The Next LLM Performance Breakthrough
By
–
Agents that natively self-orchestrate, managing their own context, tools, and sub-agents, are the next big unlock in LLM performance. Right now, a skilled engineer building an optimized harness, with thoughtful data flow, separation of concerns, sub-agent management, etc., can
-
Databricks Agents Integration with Microsoft Ecosystem and Azure
By
–
Learn more about this week's news Agent Bricks + Agent 365: https://
databricks.com/blog/bring-dat
a-intelligence-microsoft-ecosystem-agent-bricks-agent-365
… Azure Databricks for Azure Government: https://
databricks.com/blog/databrick
s-partners-microsoft-bring-our-data-intelligence-platform-azure-government
… AI/BI Genie with Copilot Studio + Microsoft Foundry: https://
techcommunity.microsoft.com/blog/analytics
onazure/azure-databricks-genie-integration-with-copilot-studio-and-microsoft-foundry-is-/4471087
… SAP Databricks on Azure: https://
ignite.microsoft.com/en-US/sessions
/BRK136?source=sessions
… -
Databricks Microsoft Partner AI Agents Enterprise Ready
By
–
A big week for Azure Databricks at #MSIgnite!
— Databricks (@databricks) 21 novembre 2025
Watch Databricks Co-founder and CEO, @alighodsi, and @Microsoft EVP of Cloud+AI, @scottgu, share how we’re partnering to make AI simple, governed and enterprise-ready.
– Agent Bricks + Agent 365: Build AI agents that operate across… pic.twitter.com/JRnCNAbtDqA big week for Azure Databricks at #MSIgnite! Watch Databricks Co-founder and CEO, @alighodsi
, and @Microsoft EVP of Cloud+AI, @scottgu
, share how we’re partnering to make AI simple, governed and enterprise-ready. – Agent Bricks + Agent 365: Build AI agents that operate across -
Prompt takes screenshots, adapts to new design, tracks progress for full website
By
–
this is cool. this prompt also can:
— God of Prompt (@godofprompt) 21 novembre 2025
1/ take your screenshots
2/ adapt them to new design
3/ track progress to build full website https://t.co/satdzGzAQJ pic.twitter.com/ttEPbk1KPPthis is cool. this prompt also can: 1/ take your screenshots
2/ adapt them to new design
3/ track progress to build full website -
Prometheus Seeks Founding CEO for Autonomous Drone Swarm AI
By
–
We’re looking for a Founding CEO / Co-founder for Prometheus, a new company we are spinning out from GoodAI. What Prometheus is: A next-generation autonomous drone swarm system for GPS-denied and signal-denied environments. ✦ LLM-driven reasoning / multi-modal / onboard ✦
-
Abacus.AI DeepAgent Listener: Real-time AI Meeting Assistant
By
–
Try out our https://t.co/ZIrQtI6s1j DeepAgent Listener!
— Abacus.AI (@abacusai) 21 novembre 2025
It’s a real-time AI assistant and listener that can provide:
•Live meeting notes
•Replies to tough questions
•Real-time insights pic.twitter.com/fb583JqNxyTry out our http://
Abacus.AI DeepAgent Listener! It’s a real-time AI assistant and listener that can provide:
•Live meeting notes
•Replies to tough questions
•Real-time insights