These steps should be part of a visible, auditable plan, an “auditable artifact”, as Karpathy says, that helps users and builders understand exactly how the agent tackled the work. 6/6
ETHICS
-
AI Control: Putting Artificial Intelligence on a Leash
By
–
The solution? In the words of Karpathy, we need to “put AI on a leash.” 4/6
-
Karpathy warns against excessive enthusiasm for autonomous AI agents
By
–
Andrej Karpathy recently warned we're getting "way too excited" about autonomous AI agents. “If I’m just vibe coding AI is great, but if I’m trying to really get work done, it’s not so great to have overreactive agents.” 2/6
-
Karpathy’s AI Leash: Building Enterprise Trust in AI Systems
By
–
Karpathy’s leash isn’t a shackle, it’s how enterprises learn to trust AI. Do you agree @karpathy ? We explore Karpathy’s idea of “putting AI on a leash” here: https://
ai21.com/blog/karpathys
-leash/
… 1/6 -
E/acc Movement Gains Influence Over Effective Altruism in AI Policy
By
–
Had to pause reading America's AI Action Plan when I saw e/acc having an impact. Would this have happened if the effective altruists had the influence they expected? Not likely. Thanks again to @beffjezos for making it fun to dunk on doomers everywhere. Now back to
-
Deepfakes and synthetic media in legal systems
By
–
I saw "Combat Synthetic Media in the Legal System" and briefly thought it might be about lawyers citing hallucinated cases, but it's about deepfakes
-
Why AI Labs Delay Model Releases: Safety and Strategy
By
–
You can interpret any AI lab’s delay in model or system release to be due to safety evals, risk concerns, production gaps, rollout friction, leadership hesitancy, or strategic value hoarding. Could be one of these, could be many. Worth keeping an open mind about causes.
-
ICML 2025 Publication Ethics Guidelines and Standards
By
–
From ICML’s Publication Ethics page: https://
icml.cc/Conferences/20
25/PublicationEthics
… -
ICML Statement on Hidden Subversive LLM Prompts
By
–
ICML’s Statement about subversive hidden LLM prompts We live in a weird timeline…
-
Inverse Scaling: Longer Reasoning Steps Decrease Model Accuracy
By
–
letting models think longer can decrease, not improve, accuracy "Inverse Scaling in Test-Time Compute" this paper shows that increasing inference steps produces an “inverse‑scaling” effect, longer CoT amplify distraction, framing over‑fit, deductive focus loss & more