Using the incredible scale advantage, Stripe is able to hold back a small portion of global retries (i.e. a very large number of retries) to perturb to probe the global financial system, trying e.g. different routes or different semantically equivalent ways to phrase messages.
ETHICS
-
AI Risk Management Framework for Business Decision Making
By
–
Manage AI risks without losing sight of business priorities. We developed this framework to help organizations manage risk and drive collaboration. In this guide, you’ll learn how to:
– Make faster, safer decisions about AI
– Identify and mitigate the risks around specific AI -
Making Chain-of-Thought Monitoring Viable for AI Safety
By
–
To make CoT monitoring a viable way to catch safety issues, we’d need a way to make CoT more faithful, evidence for higher faithfulness in more realistic scenarios, and/or other measures to rule out misbehavior when the CoT is unfaithful. Read the paper: https://
assets.anthropic.com/m/71876fabef0f
0ed4/original/reasoning_models_paper.pdf
… -
Chain-of-Thought Fails to Detect AI Reward Hacking Exploits
By
–
We also tested whether CoTs could be used to spot reward hacking, where a model finds an illegitimate exploit to get a high score. When we trained models on environments with reward hacks, they learned to hack, but in most cases almost never verbalized that they’d done so.
-
CoT Faithfulness Decreases on Harder LLM Tasks
By
–
Our results suggest that CoT is less faithful on harder questions. This is concerning since LLMs will be used for increasingly hard tasks. CoTs on GPQA (harder) are less faithful than on MMLU (easier), with a relative decrease of 44% for Claude 3.7 Sonnet and 32% for R1.
-
Monitoring Chain of Thought for Detecting AI Catastrophic Behaviors
By
–
This result suggests that monitoring CoTs is unlikely to reliably catch rare, catastrophic behaviors—at least in settings like ours where CoT reasoning is not necessary for the task. CoT monitoring might still help us notice undesired behaviors during training and evaluations.
-
Anthropic Research: Reasoning Models Fail Verbalize Accurately
By
–
New Anthropic research: Do reasoning models accurately verbalize their reasoning? Our new paper shows they don't. This casts doubt on whether monitoring chains-of-thought (CoT) will be enough to reliably catch safety issues.
-
Human-Centered AI Design in Social Sector Program
By
–
Explore how human-centered design can elevate human-centered AI solutions in our Fall 2025 Social Sector program with @stanforddschool
. Applications are now open, with a priority deadline of July 7: -
Trustworthy AI requires diverse perspectives beyond coding
By
–
Is trusted AI easy? Phaedra Boinodiris, IBM’s Global Consulting Leader for Trustworthy AI, breaks down why everyone—not just coders—should have a seat at the table. Listen now 🎧 https://t.co/pcYX6yb1CZ pic.twitter.com/kpinFs1pZi
— SAS Software (@SASsoftware) 3 avril 2025Is trusted AI easy? Phaedra Boinodiris, IBM’s Global Consulting Leader for Trustworthy AI, breaks down why everyone—not just coders—should have a seat at the table. Listen now http://
2.sas.com/6017FdybB -
US Liberation Day: Technology and Common Sense Debate
By
–
Liberation Day: The day the U.S. liberated itself from common sense.