Tech giant Anthropic has hired a philosopher to teach its artificial intelligence the difference between right and wrong amid recent reports of immoral behavior by its chatbots. Amanda Askell, the in-house thinker assigned with the responsibility of teaching Claude AI, the ethics, principles, wisdom, knowledge, and existence, shaping its moral compass.
Amanda is a philosopher by profession and works at Anthropic, examining where thinking misfires and guides it towards responsible behaviour.
Anthropic is the most closely watched AI model, used most by tech minds focused on making their AI models smarter and more human-like.
The company is training its Large Language Models (LLMs) on vast amounts of data, but to make them ‘safer for’humans’, Ahumans’c has now decided to teach its AI ethics as well, and that job is given to Amanda and her team.
Teach morals to Claude.
Amanda teaches Claude morals by studying patterns through engaging in its conversations and testing how it responds to ethically complex scenarios.
“I’m a phi working on fine-tuning and AI alignment at Anthropic. My team trains models to be more honest and to have good character traits. It works on developing new fine-tuning techniques so that our interventions can scale to more capable modelsreads hishis official bio on the company’s website.
According to a report by The Wall Street Journal, Amanda and her team wanted to make the chatbot polite and moral, and to develop a sense of identity. The goal is to ensure that Claude understands itself as a helpful and humane assistant, rather than a system that works on command.
Helping Anthropic build boundaries around safety
In simple terms, she is helping Anthropic set boundaries for safety so that, in the future, AI does not veer into unethical territory and harm humans.
Before joining Anthropic, Amanda worked as a research scientist on the policy team at OpenAI, where she focused on AI safeguards through debate and human baselines for accurate AI performance.
It was found that Anthropic, a Philosopher, is to be appointed amid growing scrutiny of AI systems. As the recent Grok controversy, it invaded the privacy of women and generated explicit images of women and young girls.
Chatbots, be they Claude or rivals like Google, have been tested in situations where users formed intense emotional attachments or received problematic advice.
As the company recently released a study examining potential loopholes in its models, including concerns about whether advanced systems like Claude Opus 4.6 could be missed in harmful ways.
